This Appendix defines terms used in the HiRDB manuals.

abstract data type

An abstract data type is a unique data type defined by the user so that data with a complex structure and operations involving it can be handled in SQL.

abstract data type column structure base table

A table containing abstract data type columns and from which the abstract data type columns have been removed is called an abstract data type column structure base table.

accepting unit

A unit in which is located a guest BES when the standby-less system switchover (effects distributed) facility is used.

access privilege

An access privilege is a permission that enables you to access a table; access privileges are set on a table-by-table basis. There are four access privileges:

• SELECT privilege
• INSERT privilege
• DELETE privilege
• UPDATE privilege

account lock period

Period during which a consecutive certification failures account lock state is to be in effect (will be detected by the connection security facility).

ADT (Abstract Data Type)

See abstract data type.

alias IP address

A technique by which different IP addresses can share the same LAN adapter through assignment of multiple IP addresses to the LAN adapter.

all asynchronous method

One of the processing methods used in Real Time SAN Replication. When an update of a database file or system file occurs at the main site, copying to the corresponding file at the remote site is performed asynchronously.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

all synchronous method

One of the processing methods used in Real Time SAN Replication. When an update of a database file or system file occurs at the main site, copying to the corresponding file at the remote site is performed synchronously.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

alternate BES

A back-end server that takes over processing when an error occurs while the standby-less system switchover (1:1) facility is being used. Similarly, a unit containing an alternate BES is called an alternate BES unit.

asynchronous copy

One of the processing methods used to update-copy data to a remote site. Update-copying at the main site is completed without waiting for update-copying at the remote site to be completed.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

asynchronous group

A group made up entirely of asynchronous pair volumes.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

asynchronous pair volume

A pair volume that has been created by specifying async as the fence level. When data is written to the P-VOL, it is not mirrored synchronously onto the S-VOL. Thus, differences may arise between the data on the P-VOL and the data on the S-VOL, even though they are paired.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

asynchronous READ facility

When the prefetch facility is being used, the asynchronous READ facility sets up two prefetch buffers and, while the DB process is using one of the buffers, a READ process prefetches pages to the other buffer asynchronously with the DB process. By overlapping the DB processing with the prefetch input, you can reduce the processing time.

audit privilege

A privilege required for auditors. A user needs the audit privilege to perform operations such as the following:

• Loading data to audit trail tables
• Swapping audit trail files
• Searching, updating, and deleting audit trail files

audit trail

Operations performed on HiRDB such as audit commands and SQL execution records are output to a file. This record is called an audit trail, and the file to which this record is output is called an audit trail file.

audit trail table

A table used by the auditor to perform audits. It is created by loading into it information saved in an audit trail file.

auditor

A person who audits a HiRDB system. An auditor must be assigned the audit privilege. Only one auditor can be registered.

automatic log unloading facility

Normally, when system log information is unloaded by the HiRDB administrator using the pdlogunld command, it is also necessary to unload any system log file that is waiting to be unloaded. The facility that automates the process of unloading system log files is called the automatic log unloading facility.

automatic reconnect facility

When a connection to a HiRDB server is lost due to a server process going down, system switchover, network failure, or other cause, this capability automatically re-establishes the connection. By using the automatic reconnect facility, a user can continue executing the UAP without having to be aware that the connection to the HiRDB server was lost.

back-end server

A back-end server is a constituent element of a HiRDB/Parallel Server that performs database access, locking, and operational processing in accordance with execution instructions received from a front-end server.

back-end server for connecting to foreign servers

A back-end server that connects to foreign servers.

backup acquisition mode

The backup acquisition mode allows you to specify whether or not referencing and updating requests from other UAPs for an RDAREA being backed up are to be accepted while a backup is being made by the database copy utility. The backup acquisition mode is specified in the -M option of the database copy utility; there are three modes:

• Referencing/updating-impossible mode (-M x specification)
  During backup, an RDAREA being backed up cannot be referenced or updated.
  
• Referencing-permitted mode (-M r specification)
  During backup, an RDAREA being backed up can be referenced but not updated.
  
• Updatable mode (-M s specification)
  During backup, an RDAREA being backed up can be both referenced and updated.
  

backup file

A backup file stores a copy of an RDAREA that can be used to recover the RDAREA after a failure.

backup-hold

When a backup is made using a command other than the pdcopy command (backup using a function provided by another product), an RDAREA to be backed up should be placed on backup-hold. An RDAREA on backup-hold can be backed up by the backup facility of another product while HiRDB is running. To apply backup-hold to an RDAREA, specify the -b option in the pdhold command. There are four types of backup-hold, as shown in Table D-1.

Table D-1 Types of backup-hold

Types of backup-hold	Explanation
Referencing-permitted backup-hold	During backup-hold, RDAREAs that have been backed up and held can be referenced; an updating attempt will result in an SQL error (-920).
Referencing-permitted backup-hold (update WAIT mode)	During backup-hold, RDAREAs that have been backed up and held can be referenced; an updating attempt goes onto lock-release wait status until the backup-hold is released.
Updatable backup-hold	During backup-hold, RDAREAs that have been backed up and held can be referenced and updated. Even while an updating transaction is being executed, an RDAREA is placed immediately on updatable backup-hold status without placing the pdhold command on wait status.
Updatable backup-hold (WAIT mode)	During backup-hold, RDAREAs that have been backed up and held can be referenced and updated. If an updating transaction is being executed, the pdhold command is kept waiting until the transaction terminates.

The following types of backup-hold are referred to collectively as committing a database:

• Referencing-permitted backup hold
• Referencing-permitted backup hold (update WAIT mode)

base row

The row that contains the base data for all columns. If a branch row has been created, the base row contains the branch target information.

BLOB (Binary Large Object)

BLOB is large data, such as large text, image, and audio data.

BLOB and BINARY data addition update and partial extraction facility

A facility that consists of the two functions described below. Data with a defined length of 32,001 bytes or more is handled as BINARY data.

• Function for specifying a concatenation operation in the UPDATE statement's SET clause in order to add new data to registered BLOB or BINARY data
• Function for specifying the SUBSTR scalar function in order to extract only a specified portion of BLOB or BINARY data

This facility can reduce the memory requirement to the amount of BLOB or BINARY data that is to be added or extracted.

BLOB data file output facility

With this facility, instead of retrieved BLOB data being returned to the client, it is output to a file in a unit that contains either a single server or a front-end server, and only the name of the file is returned to the client. Use of this facility prevents large increases in memory use during data retrieval operations.

block transfer facility

The block transfer facility transfers data from any number of rows from a HiRDB server to a HiRDB client.

branch row

Data is partitioned onto another page when its length exceeds one page. Rows that are stored on the other page in this manner are called branch rows.

changing the partition storage conditions

Use of ALTER TABLE to change the partition storage conditions for a key range partitioned table. By changing a table's partition storage conditions, RDAREAs containing old data can be reused, thereby saving time.

Note:

You can use ALTER TABLE to change the partition storage conditions of tables that have been partitioned using one of the following methods:

Boundary value specification

Storage condition specification (when the equal sign (=) was used as the storage condition comparison operator)

character code set

The sets of character codes that can be used in the UNIX version of HiRDB and in the Windows version of HiRDB are somewhat different. The following table lists the character code sets that can be used on each platform:

Character code set	UNIX version	Windows version
sjis (Shift JIS Kanji Code)	Y	Y
ujis (EUC Japanese Kanji Code)	Y	N
chinese (EUC Chinese Kanji Code)	Y	Y
utf-8 (Unicode)	Y	Y
lang-c (Single Byte Character Code	Y	Y

Legend:

Y: Can be used

N: Cannot be used

check constraint

A constraint used to maintain data integrity in tables by checking specified constraint conditions when data is inserted or updated and suppressing operations on data that does not meet the conditions.

check pending status

HiRDB limits the data operations that can be performed on referencing tables and check constraint tables when the integrity of table data can no longer be guaranteed due to execution of SQL code or a utility on a table in which a referential constraint or a check constraint is defined. This status indicates such a state, in which data operations have been limited because data integrity cannot be guaranteed.

closing RDAREAs

Process of prohibiting accesses from HiRDB to RDAREAs (closing the RDAREAs). To execute a utility that re-creates RDAREAs, the RDAREAs must first be closed.

When an RDAREA is closed, some RDAREA definition information retained in memory by HiRDB and data storage information are discarded from memory and must be re-created the next time the RDAREA is opened.

For example, when pdmod is to be used to re-initialize or delete RDAREAs or pdrstr is to be used to recover a database, the RDAREAs must first be closed.

To close an RDAREA whose open attribute is not SCHEDULE, the RDAREA must be shut down before it can be closed.

cluster key

A cluster key is a table column that is specified as the key for storing rows in ascending or descending order of the values in the specified column.

compiling

Compiling is the process of converting a post-source program created by an SQL preprocessor in order to create a machine language program (object code).

concurrent connections, maximum number of

The number of connections that can be established at the same time to a HiRDB server (single front-end server in a HiRDB/Parallel Server system); this value is specified in the pd_max_users operand. Connection requests that exceed the maximum number of concurrent connections result in a CONNECT error.

connection frame guarantee facility for client groups

Facility that groups clients connected to HiRDB and guarantees the maximum number of HiRDB connections allowed for each client group.

connection security facility

Facility for enhancing password security. This facility enables the user to set a minimum size for passwords (in bytes) in order to prohibit use of simple or easily-guessed passwords. It can also set a maximum number of consecutive unsuccessful attempts to enter a password (consecutive certification failures).

consecutive certification failure account lock state

If user authentication fails because of too many consecutive failures to enter the correct password, that user is placed in consecutive certification failure account lock state. A user in this status can no longer establish connection with HiRDB (CONNECT).

consistency group

A group that guarantees maintenance of the update order integrity of S-VOLs. Integrity is maintained between the order in which data is written on the P-VOL and the order in which data is updated to the S-VOL on all asynchronous pair volumes that make up the consistency group.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

constructor function

The constructor function is used to generate an abstract data type value.

cursor

A cursor is used during data retrieval or updating to maintain the most recent extraction position, in order to extract multiple rows of a retrieval result one row at a time. DECLARE CURSOR is used to declare a cursor, the OPEN statement is used to open a cursor, the FETCH statement is used to retrieve search results and advance the cursor to the next row, and the CLOSE statement is used to close a cursor.

database

A database is a system that organizes systematically tables, indexes, user-defined types, stored procedures, stored functions, and access privileges that are created for processing jobs.

data definition language

A language for defining the structure and contents of a database.

data dictionary

A data dictionary stores database design specifications, including database table structures, column definitions, and index definitions. In the case of a HiRDB/Single Server, the single server provides centralized management of the data dictionary. With a HiRDB/Parallel Server, centralized management of the data dictionary is provided by the dictionary server.

data dictionary LOB RDAREA

A data dictionary LOB RDAREA stores either stored procedures or stored functions. The two data dictionary LOB RDAREAs are one for storing the definition sources and one for storing the objects of stored procedures or stored functions.

data directory RDAREA

A data directory RDAREA stores dictionary tables for managing the analysis results of a definition SQL and the dictionary table indexes.

data loading

Data loading is the process of storing data in a table. Data loading is performed by the database load utility (pdload).

data loading with synchronization points set

Normally, while data is being loaded, transactions cannot be reconciled until the storage processing of all the data has been completed. This means that synchronization point dumps cannot be obtained during execution of the database load utility. If HiRDB terminates abnormally during loading of a large quantity of data, it will take a long time to restart HiRDB. To resolve this problem, you can set synchronization points for any number of data items during data loading in order to reconcile transactions. This is called data loading with synchronization points set.

To perform data loading with synchronization points set, you must specify a synchronization point lines count, which is the number data items to be stored before a synchronization point is set. This value is specified in the option statement of the database load utility. When a large amount of data is to be loaded into a table, you should consider whether or not to execute data loading with synchronization points set.

Synchronization points can also be specified in the database reorganization utility; this is called reorganization with synchronization points set.

data manipulation language

A language for specifying database operations so that a business application program can perform operations on the data in a database.

data warehouse

A data warehouse is the database system concept that enables databases residing at a mainframe, UNIX servers, and PC servers to be linked, so that they can be accessed in a manner transparent to the various different operating systems. With a data warehouse, end-users can access a database from PCs and the system administrator can access it by means of a system analysis tool, so that the data warehouse provides to each the necessary database management functions. The replication functions (HiRDB/Datareplicator and HiRDB Dataextractor) that can be used in data warehouse databases are available in HiRDB.

deadlock

Deadlock occurs when multiple transactions are competing for the same resources and are waiting for each other to release their resources.

default constructor function

The default constructor function is created automatically by HiRDB and identified with the same name as an abstract data type; the default constructor function does not take any arguments.

default RD node

The HiRDB system assumed as the connection target in a client environment definition (when using the CONNECT statement to connect without specifying an RD node name). DBMS nodes other than HiRDB are called distributed RD nodes.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

deferred write processing

Deferred write processing writes pages that have been updated in the global buffer onto the disk when the number of updated pages reaches a certain value, rather than when a COMMIT statement is issued. The point at which the number of updated pages reaches the certain value (as determined by HiRDB) is called the deferred write trigger.

delayed batch creation of plug-in index

Instead of adding plug-in index data when row data is added to a table, the plug-in index delayed batch creation facility can be used to add the plug-in index data in a batch later with the database reorganization utility.

This facility is used for adding (or updating) a large volume of row data to a table for which a plug-in index is defined.

delayed rerun

Delayed rerun is the mode that enables rollback and acceptance of new transactions to be started simultaneously during system recovery.

dictionary server

The server that centrally manages data dictionaries that contain database definition information is called the dictionary server.

differential backup facility

The differential backup facility backs up only the information that has changed since the previous backup was made, which reduces the amount of time required to make a backup. Consider using the differential backup facility when the database is large and the amount of data updating has been relatively small.

differential backup management file

A file required to use the differential backup facility. A differential backup management file stores information obtained when differential backups are made. HiRDB uses this file when backups are made and during recovery of a database using backups.

Directory Server linkage facility

A facility (such as Sun Java System Directory Server) that is used to manage and authenticate HiRDB users. This capability is provided by a process called the Directory Server linkage facility. By using a directory server, you can centrally manage organizational and user information (user IDs, passwords, affiliations, and job titles) that is otherwise managed separately in HiRDB, Groupmax, and other systems.

distributed client

A distributed client is a source of requests to access a remote database (a node with a DBMS that a UAP accesses directly).

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

distributed client facility

A facility that enables the HiRDB system on the local node to function as a distributed client, thus providing remote database access to a distributed server.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

distributed database facility

Use of a remote database access facility provided by the DF/UX distributed database products allows the user to access other database systems to retrieve and update data. There are two remote database access facilities:

• Distributed client facility
• Distributed server facility

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

distributed nest-loop-join

A distributed join in which nested-type loop processing is performed on retrieved rows that match a join condition: more specifically, nested-type processing is conducted by retrieving each row from a particular table, issuing an SQL statement for each row, and matching up the row to the other row of the counterpart table residing on a foreign server.

distributed RD node

RD nodes that are not default RD nodes. Nodes on DBMSs other than HiRDB are referred to as distributed RD nodes.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

distributed server

A distributed server is a node (with a database) that accepts requests from distributed clients and performs operations on the database.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

distributed server facility

A facility that enables the HiRDB system on the local node to function as a distributed server, so that it can accept requests from distributed clients and perform database operations.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

dynamic SQL

A method of generating SQL while a UAP is executing. Contrast with static SQL, which is a method of coding SQL within a program when a UAP is created.

dynamic update of global buffer

Dynamic update of global buffer refers to the ability to add, change, and delete global buffers with the pdbufmod command while HiRDB is running.

embedded UAP

An embedded UAP is a UAP in which an SQL is written directly into a source program that is written in a high-level language (C or COBOL).

encapsulation

The process of separating an abstract data type interface from its implementation method and keeping this implementation method hidden from the user is called encapsulation.

exceptional key value

An exceptional key value is a key value that is to be excluded in order not to create a useless index key during index definition. If an index has the null value in all rows of the column or has key values that are duplicated multiple times, the null value can be specified as the exceptional key value.

facility for conversion to a decimal signed normalized number

The facility for conversion to a decimal signed normalized number converts the sign part of signed packed-format data (decimal, date interval, time interval) that has been input; the conversion is performed in accordance with the rules described in Section 7.7 Facility for conversion to a decimal signed normalized number. The process of converting a sign is called normalization of the sign part.

facility for monitoring abnormal process terminations

A facility that monitors the number of times a server process terminates abnormally over a given period of time and terminates HiRDB (or the unit in the case of a HiRDB/Parallel Server) abnormally when the number of server abnormal terminations exceeds the value specified in the pd_down_watch_proc operand. You should use this facility in conjunction with the system switchover facility. Because this facility terminates HiRDB abnormally when a specified number of server process abnormal terminations occurs, the system switchover facility will provide rapid system switchover. If you do not use this facility, the system will not be switched over in such a situation because HiRDB will not terminate abnormally.

facility for parallel writes in deferred write processing

A facility for performing deferred write processing by executing multiple processes in parallel, which reduces the amount of time required for write processing.

facility for predicting reorganization time

Typically, the user must determine the necessity for and timing of table or index reorganization or RDAREA expansion on the basis of the results of executing the pddbst command. This sometimes results in tables being reorganized unnecessarily or in tables not being reorganized when needed because of human error (such as failure to read messages that have been output).

To simplify these operations, the facility for predicting reorganization time enables HiRDB to predict when reorganization should be performed.

facility using arrays

Array-type variables enable multiple processes to be executed by a single execution of an SQL statement. Use of this facility reduces the number of communications between HiRDB client and server.

The facility using arrays can be applied to the FETCH, INSERT, UPDATE, and DELETE statements.

falsification prevention facility

Facility that prevents updating of table data other than the data in updatable columns in order to protect table data from errors or unauthorized tampering. Tables on which the falsification prevention facility is applied are called falsification-prevented tables. When the falsification prevention option is specified, the only operations permitted on the table are row addition, retrieval, deletion of data whose deletion-prevented duration has expired, and updating of updatable columns. A table definition can also be changed to convert an existing falsification-unprevented table into a falsification-prevented table. If no deletion-prevented duration is specified, the table and its data cannot ever be deleted.

fence level

A pair volume copying method or a level for guaranteeing the integrity of data on pair volumes when an update-copy error occurs. The fence levels are data, never, and async. For details about the levels, see the RAID Manager documentation.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

FIX attribute

FIX is the attribute assigned to a table whose row length is fixed.

FIX attribute table

A table whose row length is fixed.

FIX hash partitioning

FIX hash partitioning is a method of row-partitioning a table using a hash function that divides the table by row and stores the values of the columns that comprise the table evenly among RDAREAs. The column specified as the condition for row partitioning of the table is called the partitioning key. When FIX hash partitioning is used, HiRDB recognizes the portion of the table that is partitioned into each RDAREA. Consequently, during retrieval processing, only those back-end servers in which the applicable data is presumed to be located must become retrieval targets.

flexible hash partitioning

Flexible hash partitioning is a method of row-partitioning a table using a hash function and dividing the values of the columns that comprise the table evenly among RDAREAs. The column specified as the condition for row partitioning of the table is called the partitioning key. When flexible hash partitioning is used for partitioning and storage of a table in RDAREAs, it is not possible to tell which portion of the table is in which RDAREA. Consequently, during retrieval processing, all back-end servers in which the applicable table is located must become retrieval targets.

floatable server

A floatable server is a back-end server that is dedicated to sort and merge operations that involve a heavy processing load; it is not used to access table data. Use of floatable servers improves the overall performance of HiRDB. Table data is not stored in a back-end server that is assigned as a floatable server. A back-end server that does not perform data accesses due to the nature of a particular transaction can be used temporarily as a floatable server. A floatable server can be set up only on a HiRDB/Parallel Server.

foreign index

Definition information of an index that is defined for a table in a foreign database that is mapped to a foreign table. Foreign indexes are used during optimization. The actual index is not on HiRDB. It can be defined when you are using the HiRDB External Data Access facility.

foreign key

One or more columns defined for a referencing table. The value of a foreign key is restricted in such a manner that it takes either the same value as the primary key that references it or a value that includes the null value.

foreign server

An external DBMS that is accessed from HiRDB by using the HiRDB External Data Access facility. Once a foreign server has been defined in HiRDB, HiRDB can access foreign tables.

foreign table

A table defined in HiRDB based on the definition information for a table on a foreign server. This table is needed to access tables on foreign servers. Note that HiRDB manages only the definition information for the table; the foreign server manages the actual table. It can be defined when you are using the HiRDB External Data Access facility.

free page

A page in which no data is stored. A free page may be either a used free page or an unused free page.

free page reuse mode

A method of searching for free space in an RDAREA when HiRDB is saving data. When a segment becomes full, a search is conducted to find free space in used pages before an unused segment is allocated. The next search start position is also remembered, so that, the next time, searching for free space begins at this position.

free segment

A segment in which no data is stored. A free segment may be either a used free segment or an unused free segment.

free space reusage facility

A facility by which free space in used segments is utilized once the number of segments specified by the user is reached and the last segment becomes full. This is done by switching the page search mode to the free page reuse mode. Once the free space in all the segments of the specified number of segments runs out, the free space reuse facility activates the new page allocate mode, and a new unused segment is allocated.

front-end server

A front-end server is a constituent element of a HiRDB/Parallel Server that determines the method of database access to be used from an executed SQL and provides execution instructions to back-end servers.

global buffer

A buffer that is used for input and output of data stored in RDAREAs on disk. Global buffers are allocated in common memory. Global buffers are always allocated to an RDAREA or an index.

A LOB global buffer is allocated in the following cases:

• When a plug-in index is being stored
• When benefits can be expected from buffering because there is a relatively small amount of data
• When LOB data that will be accessed frequently is being stored

global buffer pre-writing

A facility that writes specified tables and indexes into global buffers in advance. This improves the buffer hit rate.

global deadlock

A form of deadlock that occurs between servers in a HiRDB/Parallel Server configuration.

guest area

Resources for a back-end server that is associated with a guest BES.

guest BES

A back-end server whose processing has been moved to another unit by the standby-less system switchover (effects distributed) facility in the aftermath of an error. The unit for a guest BES is called the accepting unit.

HA monitor

A cluster software program for implementing a system switchover facility. It is packaged with the Hitachi HA Toolkit Extension.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

hash facility for hash row partitioning

The addition of RDAREAs to accommodate an increase in the amount of data in a hash-partitioned table (increasing the number of row partitions in a table) can create an imbalance in the amount of data in the existing RDAREAs and in the new RDAREAs. The hash facility for hash row partitioning can correct any such imbalance resulting from an increase in the number of table row partitions.

The hash facility for hash row partitioning can be applied to both FIX hashing and flexible hashing.

heterogeneous system configuration

Typically with a HiRDB/Parallel Server, all units must be based on the same platform. However, if certain conditions are satisfied, a HiRDB/Parallel Server system made up of units on different platforms can be configured. This type of configuration is called a heterogeneous system configuration.

HiRDB administrator

Of the users who log onto the operating system using a system administrator user ID, the HiRDB administrator is the user who has the privilege to manipulate HiRDB itself. The HiRDB administrator's privilege permits execution of HiRDB commands and conveys ownership of HiRDB directories and files.

HiRDB client

A workstation or personal computer on which HiRDB/Developer's Kit or HiRDB/Run Time has been installed.

HiRDB External Data Access Adapter

A product that handles the interface differences with a foreign server (an external DBMS). The following HiRDB External Data Access Adapters are available:

• HiRDB External Data Access Adapter for HiRDB and XDM/RD E2
• HiRDB External Data Access Adapter for ORACLE

The HiRDB External Data Access Adapter functionality for HiRDB and XDM/RD E2 is included in HiRDB External Data Access.

HiRDB External Data Access facility

A facility for accessing external DBMSs (ORACLE, and so on) from a HiRDB interface. This facility provides the following benefits:

• Enables you to view and update in a single table information from databases built with multiple DBMSs residing at multiple locations.
• Even when there are multiple environments on which the DBMSs are running, you can access tables from the HiRDB interface.

HiRDB file

A file that is used by HiRDB. A HiRDB file is composed of contiguous segments in a HiRDB file system area. It stores tables, indexes, and information necessary to recover the status of the system in the event of an error.

HiRDB file system area

An area in which HiRDB files are created. Separate HiRDB file system areas are created for different application purposes, for system files, for work table files, and for RDAREA files.

HiRDB.ini file

A file that contains information required by utilities that execute on HiRDB clients and for the HiRDB SQL Executer to connect to a HiRDB server. This file must be available on both the server PC and the client PC.

HiRDB.NET data provider

Data provider required in order to access databases from applications compatible with ADO.NET. The HiRDB.NET data provider complies with the ADO.NET specifications.

HiRDB system definition files

The HiRDB system definition files store the HiRDB system definition information that is used to determine the HiRDB operation environment (system configuration, control information, and each server's execution environment).

host BES

A back-end server that is defined on an applicable unit by the standby-less system switchover (effects distributed) facility. The unit for a host BES is called a regular unit.

Hub optimization information definition

A definition file that specifies whether or not queries that target a foreign server are to be executed on the foreign server and that provides cost parameters related to optimization. Use of this definition file is optional, but it is recommended that the provided sample file be used without modification.

hybrid method

One of the processing methods used in Real Time SAN Replication. When an update of a system file at the main site occurs, copying to the corresponding file at the remote site is performed synchronously. When an update of a database file at the main site occurs, copying to the corresponding file at the remote site is performed asynchronously.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

incorporation during commit

Incorporation during commit is the process of writing pages that have been updated in the global buffer onto the disk when a COMMIT statement is issued.

index

An index is assigned to a column as a key for retrieving the table. An index consists of a key and key values. The key is the column name that describes the contents of the column, and the key values are the actual values in the column.

An index can be either a single-column index or a multicolumn index. A single-column index is based on a single column of a table; a multicolumn index is based on more than one column of a table.

index page split

HiRDB indexes have a B-tree structure. Consequently, an index page split occurs when there is no free area available in an index page to which a key is to be added. An index page split occurs when HiRDB splits the index information of an index page in order to allocate a free area and moves the second half of this information to a new page.

index reorganization

Rearrangement of an index based on information derived by retrieving index key information to create an index information file. Index reorganization can be performed by index or by index RDAREA.

Over time, deleting (DELETE) and updating (UPDATE) data degrades an index's storage efficiency, which reduces performance when the index is used for searches. Index reorganization is performed to minimize such performance degradation.

inheritance

Inheritance is the process by which a lower-order abstract data type inherits attributes and routines defined in a higher-order abstract data type.

inheriting a database

The operations after a site switchover has occurred that involve shutting down HiRDB after the log application processing has been completed and the site status has been changed. Inheriting a database is performed with the pdrisedbto command.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

inner replica facility

A facility by which duplicate RDAREAs (replica RDAREAs) are defined and manipulated. Using a disk system or software that provides mirroring, the inner replica facility allows you to access the duplicated database. For details about the inner replica facility, see the HiRDB Staticizer Option Version 7 Description and User's Guide.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

integrity constraint

An integrity constraint is a constraint that is available for guaranteeing the validity of data in a database.

There are two integrity constraints:

• NOT NULL constraint (does not permit the null value to be set in a specified column)
• Uniqueness constraint (requires that the values in a specified column be unique and does not permit value duplication in the column)

interface area

An area used for exchanging information between HiRDB and UAPs. HiRDB provides the following seven interface areas:

• SQL Communications Area
• SQL descriptor area
• Column Name Descriptor Area (SQLCNDA)
• Type Name Descriptor Area (SQLTNDA)
• Embedded variables
• Indicator variables
• Parameters

inter-process memory communication facility

When a HiRDB server and a HiRDB client are running on the same server machine, memory is used for inter-process communication to increase communication speed between the HiRDB server and the HiRDB client. This functionality is provided by the inter-process memory communication facility, which is used by specifying MEMORY in the PDIPC operand of the client environment definition.

IP address

An address used by the IP protocol.

Java stored function

A Java stored function is a stored function in which procedures are coded in Java.

Java stored procedure

A Java stored procedure is a stored procedure in which procedures are coded in Java.

JDBC driver

A driver created for HiRDB based on the interface defined in the JDBC standard. The JDBC driver is needed to execute Java stored procedures and Java stored functions. It is installed by selecting it when you install a HiRDB client.

join

Processing that combines two or more tables.

JP1

JP1 is a series of products that provide functions for batch job operations, automatic system operation, document output control, and file backup.

JP1 is used to automate system operations and reduce labor costs.

key range partitioning

Key range partitioning is a method of table partitioning that divides a table into groups of rows, using ranges of values in a column of the table as the partitioning condition. The column specified as the condition for row partitioning is called the partitioning key.

LAN adapter

Hardware that converts data for connecting a computer to a LAN.

language compiler

A program that converts a post-source program created by an SQL preprocessor into a machine-language program (object).

large file

A HiRDB file system area of at least 2 gigabytes in size is called a large file. Normally, the maximum size of a HiRDB file system area is 2047 megabytes (approximately 2 gigabytes). A HiRDB file system area whose size needs to exceed this limit must be created as a large-file HiRDB file system area. The maximum size of a HiRDB file system area created as a large file is as follows:

• HP-UX version: 131,071 megabytes
• Solaris, AIX 5L, or Linux version: 1,048,575 megabytes

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

linkage

The process of creating a single executable program file (load module) by combining the object modules created by a language compiler.

linker

A program that combines the object modules created by a language compiler into a single executable program file (load module).

list

A set of data items saved under a single name (list name). Lists include sets of data items that may be saved with temporary names at intermediate stages during the process of filtering information in a series of hierarchical steps based on specification of conditions that are designed to retrieve an appropriate number of data items.

list RDAREA

Lists created by the ASSIGN LIST statement are stored in a list RDAREA. A list RDAREA is created when a narrowed search is performed.

load module

A UAP source that has been preprocessed, compiled, and linked into a single executable file.

LOB column structure base table

A table containing LOB columns from which the LOB data has been removed is called a LOB column structure base table. Data can be loaded and the database can be reorganized by LOB column structure base table or LOB column.

LOB data

Large variable object data such as documents, images, and sounds is called LOB data. LOB data is stored in user LOB RDAREAs. Data loading and database reorganization can be performed separately from the user RDAREA that stores the LOB column structure base table.

local buffer

A buffer that is used for input and output of data stored in RDAREAs on disk. Local buffers are allocated in process private memory.

locator facility

Facility that reduces a client's memory resource overload and data transfer volume during BLOB or BINARY data searches.

A locator is a 4-byte data value that is used to identify data at a server. When a locator-embedded variable is specified, such as in the INTO clause of a single-row SELECT statement or FETCH statement, the value of the locator indicating the data, not the data itself, is received as the search result. Processing that handles the data identified by a locator can be performed when a locator-embedded variable is specified in other SQL statements.

lock

A form of control managed by HiRDB to maintain database integrity.

locking

Locking is a type of control managed by HiRDB in order to maintain database integrity.

log acquisition mode

The log acquisition mode is one of the modes for acquiring a database update log during execution of a UAP or utility. This mode acquires the database update log that will be needed for rollback or rollforward when the UAP or utility is updating the contents of an RDAREA.

log application

In the log-only synchronous method, update processing of a database based on system logs copied from the transaction execution site. Log application is performed at the log application site.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

log application not possible status

The state in which integrity of the databases at the transaction execution site and the log application site is not established or the information required for log application cannot be copied correctly. If a natural disaster occurs while the system is in this status, data will be lost if the transaction execution site goes down.

When the preparations for system log application are performed while the system is in log application not possible status, the system is placed in log application possible status.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

log application possible status

The state in which integrity of the databases at the transaction execution site and the log application site is established and the information required for log application can be copied correctly. If a natural disaster occurs while the system is in this status, job processing can be transferred to the log application site without loss of data, even if the transaction execution site goes down.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

log application site

In the log-only synchronous method, the site at which update processing of databases is performed based on system logs copied from the transaction execution site. As with the transaction execution site, the log application site must be operational at all times.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

LVM

A disk device management facility that can create and manage multiple virtual devices from a set of one or more real devices. LVM enables the user to achieve disk device management with high availability and management simplicity, such as by dividing a single real device into multiple small virtual devices or by grouping multiple real devices into a single large virtual device. LVM also provides functions for improving performance and achieving redundancy.

main site

Site where the system normally used by Real Time SAN Replication is located.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

master directory RDAREA

The master directory RDAREA stores information on the RDAREAs that store dictionary tables and user-created tables and indexes; it also stores RDAREA registration location (server) information.

matrix partitioning

A method of partitioning a table by a combination of partitioning methods using two of the table columns as the partitioning key. The first column used as the partitioning key is called the first dimension partitioning column, and the second column used as the partitioning key is called the second dimension partitioning column. Matrix partitioning always involves key range partitioning with boundary values specified for the first dimension partitioning column and then partitioning the resulting data further by the second dimension partitioning column. A table partitioned by the matrix partitioning method is called a matrix partitioning table.

message queue monitoring facility

A facility that monitors for the server process not responding state. HiRDB uses the message queue during server process allocation processing. If a server process not responding state occurs, messages can no longer be fetched from the message queue. If messages cannot be fetched from the message queue after a certain amount of time has elapsed (called the message queue monitoring time), HiRDB outputs a warning message or error message (KFPS00888-W or KFPS00889-E). Output of one of these messages indicates the possibility that a server process not responding state may have occurred.

module trace

A type of troubleshooting information used for isolating errors. The executed functions and macro history of most HiRDB processes are recorded in a process-specific memory. If a process terminates abnormally and a core file is output, the contents of the process-specific memory are output to the core file, which allows a module trace to be obtained from the core file.

monitor mode

The system switchover facility can be operated in either the monitor mode or the server mode. In the monitor mode, only system failures are monitored. In the server mode, both system failures and server failures (such as HiRDB abnormal termination) are monitored. System switchover can take less time in the server mode than in the monitor mode because a standby HiRDB is started beforehand in the server mode. The system switchover facility must also be operating in the server mode to use the following functions:

• user server hot standby
• rapid system switchover facility
• standby-less system switchover facility

monitoring free area for system log file

A facility that outputs a warning message or suspends scheduling of new transactions and forcibly terminates all transactions on the server when the percentage of free space in the system log file drops below a warning value.

multi-connection facility

The multi-connection facility enables a UAP running on a HiRDB client to connect simultaneously to multiple HiRDB servers.

multi-HiRDB

Multi-HiRDB is the mode in which multiple HiRDB systems run on a single server machine.

multiple front-end servers

The configuration in which multiple front-end servers are set up is used when the CPU load for SQL processing is too heavy for a single front-end server.

narrowed search

A narrowed search is the process of narrowing in steps the records retrieved in successive searches.

A narrowed search requires creation of a list using the ASSIGN LIST statement of the data manipulation SQL. When such a list is created under appropriate conditions, its use can reduce processing time. If there are multiple conditions, multiple lists can be combined for the search.

NetBackup linkage facility

A facility designed to create backup files for use by the database copy utility (pdcopy) or database recovery utility (pdrstr) on a medium managed by NetBackup Server.

new page allocate mode

A method of searching for free space in an RDAREA when HiRDB is saving data. When a segment becomes full, first, a new unused segment is allocated. If the RDAREA runs out of unused pages, a search is then conducted to find free space in a segment already allocated to that table (in a used segment), starting at the beginning of the first used segment. Once free space is found, the data is saved to that free space.

node

In a distributed system, each communication control device making up the network, and the systems connected to these devices.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

no-log mode

The no-log mode is one of the modes for acquiring a database update log during execution of a UAP or utility. This mode does not acquire a database update log when the UAP or utility is updating the contents of an RDAREA.

normal BES

A back-end server whose processing is transferred to another server in the event of an error, through use of the standby-less system switchover (1:1) facility. Similarly, a unit containing a normal BES is called a normal BES unit.

no-split option

Normally, the data part of a variable-size character string that is at least 256 bytes in length is stored in different pages. The no-split option allows you to store one row of data in one page even when the actual data size of a variable-size character string is 256 bytes or greater. The no-split option improves the data storage efficiency. The no-split option is used by specifying NO SPLIT in the CREATE TABLE or CREATE TYPE statement.

NOT NULL constraint

The NOT NULL constraint is a limitation that does not permit the null value to be set in a column.

no transaction loss (no data loss)

With no transaction loss, update processing on transactions committed at the main site is guaranteed to be mirrored to the database at the remote site.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

null value

The null value is a special value that indicates that no value has been set.

object

A program that has been translated into machine language by a language compiler.

object relational database

An object relational database is a database that has been expanded by incorporating the concept of object orientation into a relational database model. HiRDB can handle data that has a complex structure, such as multimedia data, and the operations for such data as a unified object and can manage it in a database.

operation releasing check of unload status

A system log file can become a swap target only if it is in one of the following two statuses:

• Overwriting enabled status
• Extraction completed status (HiRDB Datareplicator)

The unloading status is unrelated to the statuses required in order for a system log file to be a swap target. Performing this operation eliminates the need to unload or release system log files.

operation unloading system log

The operation of unloading system logs. When a database is recovered, unload logs that have been unloaded from the system logs are used as the input to the database recovery utility.

operation without unloading system log

Operating mode in which system log files are not unloaded. When a database is recovered, system logs are entered directly as input information to the database recovery utility, without unload logs being used.

optimizing based on cost

When multiple indexes have been created for a table, HiRDB selects and uses the index that has the least access cost and that it evaluates to be optimal based on the conditions specified for the table retrieval. This process in which HiRDB selects the index it evaluates to be optimal is called optimizing based on cost.

overload

Multiple stored functions can be defined with the same name as long as their parameter counts and data types are different. Stored functions that have the same name are said to be mutually overloaded.

override

Override is the process of overwriting the definition of a lower-order abstract data type with a routine that has the same name as the routine defined for the higher-order abstract data type.

page

The page is one of the data storage units; it is the smallest unit for database input/output operations. There are three types of pages:

• Data page: Page for storing rows of a table
• Index page: Page for storing index key values
• Directory page: Page for storing RDAREA status management information

pair logical volume group

A group of pair logical volumes. Pair logical volumes on which files have been distributed are handled in units of pair logical volume groups.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

pair logical volumes

Volumes paired between servers that have been logically named and configured.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

pair status

A designation that indicates the status of a pair volume. The pair statuses include PAIR, SMPL, and PSUS. For details about these statuses, see the RAID Manager documentation.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

pair volumes

With Lightning/Thunder series volumes, a volume at the main site and a volume at the remote site that correspond to each other.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

paired

A status in which a pair logical volume or pair logical volume group is paired.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

partitioning key index

An index that satisfies a particular condition becomes a partitioning key index. An index that does not satisfy the condition is called a non-partitioning key index. The condition is explained as follows.

The condition is whether the table involved is partitioned on the basis of single-column partitioning or multicolumn partitioning. When only one column is used in the table partitioning condition, the partitioning is said to be single-column partitioning; when multiple columns are used in the table partitioning condition, the partitioning is said to be multicolumn partitioning.

(a) Single-column partitioning

An index satisfying one of the following conditions is a partitioning key index:

Conditions

• Single-column index is defined for the column (partitioning key) on the basis of which storage conditions were specified when the table was row partitioned
• Multicolumn index is defined, and the column (partitioning key) on the basis of which storage conditions were specified when the table was row partitioned is the first constituent column

Figure D-1 shows an example (based on the STOCK table) of an index that becomes a partitioning key index.

Figure D-1 Partitioning key index: Single-column partitioning

Explanation

CREATE INDEX A12 ON STOCK (PCODE ASC)                 1
CREATE INDEX A12 ON STOCK (PCODE ASC, PRICE DESC)     2
CREATE INDEX A12 ON STOCK (PRICE_DESC, PCODE ASC)     3

1. If the PCODE column, which is a partitioning key, is used as an index, the index becomes a partitioning key index. If any other column is used as an index, the resulting index becomes a non-partitioning key index.
2. Specifying the PCODE column, which is the partitioning key, as the first constituent column of the index makes the resulting multicolumn index a partitioning key index.
3. Specifying the PCODE column, which is the partitioning key, as a column other than the first constituent column of the index makes the resulting multicolumn index a non-partitioning key index.

(b) Multicolumn partitioning

An index satisfying the following condition is a partitioning key index:

Condition

• Index is created on the basis of multiple columns, beginning with the partitioning key and containing all the columns specified for partitioning from the beginning and without any change in their order.

Figure D-2 shows an example (based on the STOCK table) of an index that becomes a partitioning key index.

Figure D-2 Partitioning key index: Multicolumn partitioning

Explanation

CREATE INDEX A12 ON STOCK (PCODE ASC, PRICE DESC)      1
CREATE INDEX A12 ON STOCK
            (PCODE ASC, PRICE DESC, SQUANTITY ASC)     2
CREATE INDEX A12 ON STOCK (PRICE DESC, PCODE ASC)      3
CREATE INDEX A12 ON STOCK
            (PCODE ASC, SQUANTITY DESC,PRICE DESC)     4

1. All partitioning keys (PCODE and PRICE columns) are specified, and these keys are specified in the same order as in the table definition. Therefore, this multicolumn index is a partitioning key index.
2. All partitioning keys (PCODE and PRICE columns) are specified, and these keys are specified in the same order as in the table definition. Therefore, this multicolumn index is a partitioning key index.
3. All partitioning keys (PCODE and PRICE columns) are specified, but these keys are specified in an order that differs from the table definition. Therefore, this multicolumn index is a non-partitioning key index.
4. All partitioning keys (PCODE and PRICE columns) are specified, but these keys are specified in an order that differs from the table definition. Therefore, this multicolumn index is a non-partitioning key index.

password-invalid account lock state

When limitations are set on password character strings by the connection security facility, a user violating any of the limitations is placed in password-invalid account lock state. A user in this status can no longer establish connection with HiRDB (CONNECT).

plug-in

A plug-in is a HiRDB package product that provides abstract data types in which documents, images, and other multimedia entities are defined, and a facility for rapid manipulation of complex data.

plug-in index

A plug-in index is a type of index provided by a plug-in.

POSIX library version

A HiRDB execution environment required for the Directory Server linkage facility, Java stored procedures, Java stored functions, and the HiRDB External Data Access facility. To use the POSIX library version, you specify the -l option of the pdsetup command.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

post-source

The source program generated by preprocessing embedded SQL statements.

predefined data type

A data type provided by the HiRDB system is called a predefined data type. The predefined data types include INTEGER, CHARACTER, DATE, etc.

prefetch facility

A facility that reads multiple pages at a time from a global buffer or a local buffer. It can reduce the I/O time when searches on large volumes of data are performed in HiRDB file system areas that use the raw I/O facility. It is particularly effective for searches that do not use an index, or for ascending-order searches on large data sets in tables that do use an index.

preparations for system log application

A procedure that enables performance of log application in which database integrity is established and information in files required to perform log application is set in the proper state by temporarily synchronizing the databases at the transaction execution site and the log application site.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

preprocessing

Processing performed on code before it is compiled by a language compiler. Preprocessing converts SQL statements to high-level language code.

pre-update log acquisition mode

Pre-update log acquisition mode is one of the modes for collecting a database update log when a UAP or utility executes. This mode acquires only the database update log needed for rollback when a UAP or utility is updating the contents of an RDAREA.

primary key

The primary key enables you to uniquely identify a row in a table. Columns for which the primary key is defined are subject to the uniqueness constraint and the NOT NULL constraint. The uniqueness constraint is the limitation that does not permit duplicate data in a key (a column or a set of columns)i.e., all data in the key must be unique. The NOT NULL constraint is the limitation that does not permit the null value in the key columns.

process

An instance of execution of a program. Each process is allocated a virtual space and CPU resources on a time-sharing basis. Executing more than one process in parallel (multi-processing) is one way to increase throughput.

protection mode

A protection mode can be used to select a processing method when synchronous copying to a remote site cannot be performed. A protection mode can be selected when the all synchronous method or the hybrid method is being used. The protection modes include data and never, and their processing methods are explained below:

• data: Stops update processing at the main site (update processing of volumes containing files that cannot be copied synchronously).
• never: Continues update processing at the main site.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

RAID Manager instance

A suite of facilities that identify the scope of operations and management that can be performed by RAID Manager. Each instance is identified by an instance number.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

range specification recovery

One of the methods used by the database recovery utility for recovering a database. It is a technique by which the database can be recovered to a user-specified point between the time a backup was made and the time the failure occurred. The -T option of the database recovery utility is used to specify the recovery time.

rapid grouping facility

When the GROUP BY clause of an SQL is used for grouping, sorting is performed before grouping. This is the rapid grouping facility, and it combines grouping with hashing.

rapid system switchover facility

A facility whereby server processes and the system server are started on a standby HiRDB beforehand, so that startup processing of server processes and the system server is not performed when the system is switched over. The system switchover time is reduced by the amount of time required to perform startup processing of server processes and the server system. In addition to the rapid system switchover facility, user server hot standby is available for reducing system switchover time. The rapid system switchover facility can reduce the system switchover time more than user server hot standby can (the rapid system switchover facility includes the user server hot standby functionality).

RD node

In the distributed database facility, a DBMS on a server connected on a distributed network. In the OSI-RDA protocol, it is called a resource.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

RD node name

In the distributed database facility, the node name used to specify a target server DBMS connected on a distributed network. In the OSI-RDA protocol, it is called a resource name.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

RDAREA

The RDAREA is the basic unit for storing data; an RDAREA may store 1-16 HiRDB files. Following are the types of RDAREAs:

• Master directory RDAREA
• Data dictionary RDAREA
• Data directory RDAREA
• Data dictionary LOB RDAREA
• User RDAREA
• User LOB RDAREA
• Registry RDAREA
• Registry LOB RDAREA
• List RDAREA

RDAREA automatic extension

When RDAREA automatic extension is being used and a space shortage occurs in an RDAREA, segments are added automatically to expand the capacity of the RDAREA, provided that there is adequate unused free space in the HiRDB file system area.

RDAREAs, shutting down

To limit accesses by UAPs and utilities to an RDAREA, the RDAREA can be shut down (command shutdown). If an error (such as an I/O error) occurs on an RDAREA, HiRDB may shut down the RDAREA automatically (error shutdown).

There are various shutdown modes, such as one that allows referencing and updating; the user selects the appropriate shutdown mode.

For example, when pdload is to be used to load data or pdrorg is to be used to re-organize a table, the applicable RDAREAs should first be shut down because such processing usually takes a considerable amount of time.

Real Time SAN Replication

A facility that enables jobs to continue being processed on a standby system set up at a remote site in the event of a disaster (earthquake, fire, etc.) that makes it physically impossible to quickly restore the system that is normally used.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

recovery-unnecessary front-end server

If an error occurs on the unit where the front-end server is located, transactions that were executing from that front-end server may be placed in uncomplete status for transaction determination. Because a transaction in this status has locked the database, some database referencing or updating becomes restricted. Normally, to complete a transaction that is in uncomplete status for transaction determination, the front-end server must be recovered from the error and restarted. However, if the abnormally terminated front-end server is a recovery-unnecessary front-end server, HiRDB automatically completes any transaction that was placed in uncomplete status for transaction determination. This enables other front-end and back-end servers to be used to restart database update processing. A unit containing a recovery-unnecessary front-end server is called a recovery-unnecessary front-end server unit.

reduced activation

Reduced activation is the process of starting HiRDB using only the normal units, when there are units that cannot be started. Normally, a HiRDB/Parallel Server cannot be started if any of its units cannot be started. In such a case, the reduced activation facility enables the HiRDB/Parallel Server to be started using only the remaining units even if some units cannot be activated due to errors.

reference buffer

When data is referenced, it is referenced in a global buffer. A buffer that is used for referencing data, or a buffer for a database that has been updated, is called a reference buffer.

reference-only back-end server

A back-end server in which shared tables and shared indexes of shared RDAREAs can be referenced but not updated.

referential constraint

A constraint that is defined for a specific column (foreign key) during table definition in order to maintain referential conformity in data between tables. A table for which referential constraints and a foreign key are defined is called a referencing table, and a table that is referenced from the referencing table using the foreign key is called a referenced table. For the referenced table, the primary key that is referenced by the foreign key must be defined.

registry facility

The registry facility enables HiRDB to maintain information unique to plug-ins; such information is used by the plug-ins for data manipulation.

registry LOB RDAREA

The registry LOB RDAREA stores a table for managing registry information (registry table). This RDAREA is required in order to use the registry facility. Not all plug-ins use the registry facility. The system determines automatically whether data is to be stored in the registry LOB RDAREA or the registry RDAREA on the basis of the length of the data. Stored procedures for data manipulation, such as ones that register information in the registry management table, are also stored in this RDAREA.

registry RDAREA

The registry RDAREA stores a table for managing registry information (registry table). This RDAREA is required in order to use the registry facility. Not all plug-ins use the registry facility.

regular expression

A method of representation that allows different character strings, such as any character strings, repetitions of the same character string, or any number of characters, to be expressed in a specific sequence (pattern) of characters forming part of a larger character string. Regular expressions are specified in a pattern character string of the SIMILAR predicate.

regular unit

The unit where a host BES is located when the standby-less system switchover (effects distributed) facility is used.

remote site

Site where the backup system is located for Real Time SAN Replication.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

reorganization by schema

Reorganization by schema is the process of reorganizing in a batch all the tables in a schema. To reorganize in a batch all the tables you own, you should execute the reorganization by schema by specifying the authorization identifier of the schema to be reorganized in the -t option of the database reorganization utility. The specification format is -t authorization-identifier.all.

In addition to reorganization by schema, the other units of reorganization are as follows:

• Reorganization by table
• Reorganization by RDAREA

reorganization with synchronization points set

Normally, while a table is being reorganized, transactions cannot be reconciled until the storage processing of all the data has been completed. This means that synchronization point dumps cannot be obtained during execution of the database reorganization utility. If HiRDB terminates abnormally during reorganization of a large quantity of data, it will take a long time to restart HiRDB. To resolve this problem, you can set synchronization points for any number of data items during storage of the data (reload processing) in order to reconcile transactions. This is called reorganization with synchronization points set.

To perform reorganization with synchronization points set, you must specify a synchronization point lines count, which is the number data items to be stored before a synchronization point is set. This value is specified in the option statement of the database reorganization utility. When a table containing a large quantity of data is to be reorganized, you should consider whether or not to execute reorganization with synchronization points set.

Synchronization point setting can also be specified in the database load utility; this is called data loading with synchronization points set.

repetition column

A column that consists of multiple elements in a single row is called a repetition column. An element is one of the multiple items in the repetition column row. A repetition column is defined by CREATE TABLE with the maximum number of elements specified. However, the number of elements can be increased later with ALTER TABLE. Figure D-3 shows an example of a table that contains repetition columns.

Figure D-3 Example of a table containing repetition columns

replication facility

The replication facility enhances the efficiency of data utilization in a data warehouse by linking a database in a Hitachi mainframe with a HiRDB database or a HiRDB database with another HiRDB database.

reserved word

A character string that has been registered as a keyword used in SQL statements.

Reserved words cannot be used as table names or column names. Otherwise, a reserved word can be used as a name if it is enclosed in double quotation marks (").

A keyword that has been registered as a reserved word can be deleted by the SQL reserved word deletion facility; however, some functions may become disabled if a reserved word is deleted.

rollback

Rollback is the process of invalidating the database processing performed by a transaction when an error has occurred in the transaction.

row partitioning

Row partitioning is the process of storing a table, index, or LOB column by dividing it into multiple user RDAREAs or user LOB RDAREAs. When a table is row-partitioned, its indexes can also be row partitioned in correspondence with the row-partitioned table. When a table contains a LOB column, it can be partitioned and stored in multiple user LOB RDAREAs in correspondence with the row-partitioned table. When a table is to be row-partitioned, it is necessary to specify storage conditions for the row partitioning; this is done with the CREATE TABLE statement of the definition system SQL. When an index is row-partitioned, it is necessary to specify the user RDAREAs in which the row-partitioned index is to be stored; this is done with the CREATE INDEX statement of the definition system SQL.

schema

A concept that encompasses tables, indexes, abstract data types (user-defined types), index types, stored procedures, stored functions, triggers, and access privileges.

schema definition privilege

The privilege required to define a schema.

security facility

The security facility prevents anyone who does not have the required authorization from accessing the database.

segment

The segment is a unit of data storage consisting of multiple contiguous pages. It is the unit of allocation for storage of tables and indexes. A segment can store either one table or one index.

server machine

A workstation or personal computer that runs HiRDB server software.

server mode

The system switchover facility can be operated in either monitor mode or server mode. In monitor mode, only system failures are monitored. In server mode, both system failures and server failures (such as HiRDB abnormal termination) are monitored. System switchover can take less time in server mode than in monitor mode because a standby HiRDB is started beforehand in server mode. In addition, the system switchover facility must be operating in server mode to use the following functions:

• user server hot standby
• rapid system switchover facility
• standby-less system switchover facility

shared index

An index that is stored in a shared RDAREA and can be referenced from all back-end servers. In order to achieve SQL and UAP compatibility with HiRDB/Parallel Servers, shared indexes can also be defined in HiRDB/Single Servers. If defined, however, such shared indexes are stored in user RDAREAs because shared RDAREAs cannot be defined in a HiRDB/Single Server.

shared RDAREA

A user RDAREA that can be referenced from all back-end servers. Only a HiRDB/Parallel Server can define shared RDAREAs.

shared table

A table that is stored in a shared RDAREA and can be referenced from all back-end servers. In order to achieve SQL and UAP compatibility with HiRDB/Parallel Servers, shared tables can also be defined in HiRDB/Single Servers. If defined, however, such shared tables are stored in user RDAREAs because shared RDAREAs cannot be defined in a HiRDB/Single Server.

site status

The status of a site as recognized by the log-only synchronous method. There are four statuses: initial, ready, transaction, and log application.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

skipped effective synchronization point dump monitoring facility

With the pd_spd_syncpoint_skip_limit system common definition operand, you can specify the maximum number of times synchronization point dumps can be skipped during a transaction.

If, for example, an infinite loop occurs in a UAP, synchronization point dump processing may not be performed several times in succession (processing may be skipped a number of times). If the number of times this processing is skipped exceeds a value specified by the user, the affected transaction is forcibly suspended and rolled back. This facility keeps track of the number of times in succession that synchronization dumps are skipped.

space conversion facility

The space conversion facility unifies coding of single-byte and double-byte spaces that are intermixed in table data. Double-byte spaces are coded as X'8140' in JIS Kanji code and as X'A1A1' in EUC Chinese Kanji code. Two single-byte spaces are coded as X'2020'.

SQL connection

The logical connection of a UAP to an RD node for the purpose of executing an SQL statement.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

SQL extension optimizing options

SQL extension optimizing options enable optimization of SQL at execution time by determining the most efficient access paths that can be specified, taking into consideration the status of the database.

The following SQL extension optimizing options are provided:

1. Application of optimizing mode 2 based on cost
2. Hash-execution of a hash join or a subquery
3. Suppression of foreign server execution of SQL statements that include a join
4. Forced foreign server execution of SQL statements that include a direct product
5. Suppression of unconditionally generated derived rapid search conditions that can be executed on foreign servers

SQL object

An SQL object is the object that is compiled by HiRDB when SQL statements are defined and executed.

SQL optimization options

SQL optimization options enable optimization of SQL at execution time by determining the most efficient access paths that can be specified, taking into consideration the status of the database.

The following SQL optimization options are provided:

1. Forced nest-loop-join
2. Creating multiple SQL objects
3. Increasing the target floatable servers (back-end servers for fetching data)
4. Prioritized nest-loop-join
5. Increasing the number of floatable server candidates
6. Priority of OR multiple index use
7. Group processing, ORDER BY processing, and DISTINCT set function processing at the local back-end server
8. Suppressing use of AND multiple indexes
9. Rapid grouping facility
10. Limiting the target floatable servers (back-end servers for fetching data)
11. Separating data collecting servers
12. Suppressing index use (forced table scan)
13. Forcing use of multiple indexes
14. Suppressing creation of update-SQL work tables
15. Derivation of rapid search conditions
16. Application of scalar-operation-included key conditions
17. Batch acquisition from plug-in-provided functions

SQL optimization specification

Optimizations to enhance SQL search efficiency can be specified for SQL statements. The following three SQL optimization specifications are available:

• SQL optimization specification for index utilization
• SQL optimization specification for the join method
• SQL optimization specification for the subquery execution method

SQL optimization specifications take precedence over any specifications of SQL optimization options and SQL extension optimizing options.

SQL preprocessor

A program that converts SQL statements into high-level language code so that they can be compiled by a high-level language compiler.

SQL runtime warning output facility

After SQL code is executed, HiRDB checks its runtime. If this check determines that the SQL code's runtime is greater than the time specified (set as a ratio of the PDCWAITTIME value), this facility is used to output the following warning information for that SQL code:

• SQL runtime warning information file
• Warning message (KFPA20009-W)

SQL stored function

An SQL stored function is a stored function in which procedures are coded in SQL.

SQL stored procedure

An SQL stored procedure is a stored procedure in which procedures are coded in SQL.

standby system switchover facility

A standby HiRDB separate from the HiRDB that is actively processing jobs is deployed, and if a failure occurs on the server machine or on HiRDB, job processing can be automatically switched over to the standby HiRDB. This is called the standby system switchover facility.

standby-less system switchover (effects distributed) facility

A type of standby-less system switchover facility. In the event of an error, this facility distributes to multiple active units the processing requests for back-end servers (BESs) in the erroneous unit; this is called the standby-less system switchover (effects distributed) facility.

standby-less system switchover facility

Unlike a standby system HiRDB that has been prepared for the standby system switchover facility, with the standby-less system switchover facility a standby HiRDB need not be prepared. If an error occurs, there is no system switchover to a standby HiRDB system, but another operating unit takes over the processing. This is called the standby-less system switchover facility.

The standby-less system switchover facility can be further subdivided as follows:

• Standby-less system switchover (1:1) facility
• Standby-less system switchover (effects distributed) facility

standby-less system switchover (1:1) facility

Unlike the standby system switchover facility that provides a standby HiRDB system, the standby-less system switchover facility does not require that a standby HiRDB system be kept in reserve. In the event of an error, the standby-less system switchover facility transfers processing to another active unit without switching over to a standby HiRDB system. This is called the standby-less system switchover facility.

The standby-less system switchover (1:1) facility can perform 1-to-1 unit switchover in the event of an error and assign processing to another, designated back-end server.

static SQL

A method of coding SQL within a program when a UAP is created. Contrast with dynamic SQL, which is a method of generating SQL code while a UAP is executing.

statistics log file

A file that stores statistical information (statistics logs) output by HiRDB.

status file

A file that stores system status information that may be needed to restart HiRDB is called a status file. There are two types of status files:

• Server status files
• Unit status files

status file for log application processing

A status file required by the log-only synchronous method. It is used to obtain system status information when log application processing is performed at the log application site. Its counterpart, the status file for transaction processing, is required at the transaction execution site.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

status file for transaction processing

A status file required by the log-only synchronous method. It is used when log application processing is performed at the log application site. Its counterpart, the status file for log application processing, is required at the transaction execution site.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

stored function

A facility for registering as a function in the database a data process coded in SQL or Java. Because input parameters can be assigned to a stored function and because a stored function can generate return values, a stored function can be called by specifying it as a value expression in an SQL statement. A stored function can be defined in a function in the CREATE FUNCTION or CREATE TYPE statement. Data processes coded in SQL or Java are compiled during definition, an SQL object coding of the procedure is generated, and the object, together with the definition information, is stored in the database.

stored procedure

A stored procedure enables a sequence of database access procedures coded in SQL or Java to be stored as a procedure in a database. Output or input/output parameters can be assigned to a stored procedure, and a stored procedure can be called by the CALL statement in SQL. A stored procedure can be defined in a procedure in the CREATE PROCEDURE or CREATE TYPE statement. Database operations coded in SQL or Java are compiled at definition time, an SQL object coding the access procedures is generated, and the object, together with definition information, is stored in the database.

substitutability

Substitutability is the capability to substitute a value in a lower-order abstract data type for the value in the higher-order abstract data type.

subtype

A subtype is an abstract data type that is created by customizing an abstract data type.

Sun Java System Directory Server linkage facility

By using the Sun Java System Directory Server, you can manage and authenticate HiRDB users. This capability is provided by a process called the Sun Java System Directory Server linkage facility. By using this facility, you can centrally manage organizational and user information (user IDs, passwords, affiliations, job titles) that is otherwise managed separately in HiRDB, Groupmax, and other systems.

supertype

An abstract data type that is higher in order than a specialized abstract data type (subtype) is called the supertype.

synchronization point

The point at which a transaction is completed is called a synchronization point. Synchronization point processing that validates a transaction-induced update is called a commit, and synchronization point processing that invalidates a transaction is called a rollback.

synchronization point dump file

In the event of abnormal termination of HiRDB, recovering it using system log information only will require system log information covering the entire time since HiRDB was startup, which will require a significant amount of time for recovery processing. Therefore, points can be established at fixed intervals during HiRDB operation, and required HiRDB management information can be saved at those points. This means that all system log information produced prior to the point to be used for recovery is no longer needed, thus reducing the recovery time. The file that stores the HiRDB management information obtained at each such point is called a synchronization point dump file.

synchronization point dump file for log application processing

A synchronization point dump file that is required by the log-only synchronous method. It is used to obtain synchronization points when log application processing is performed at the log application site. Its counterpart, the synchronization point dump file for transaction processing, is required at the transaction execution site.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

synchronization point dump file for transaction processing

A synchronization point dump file that is required by the log-only synchronous method. It is used when log application processing is performed at the log application site. Its counterpart, the synchronization point dump file for log application processing, is required at the transaction execution site.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

synchronous copy

One of the processing methods used to update-copy data to a remote site. After update-copy processing at the remote site is completed, update-copy processing at the main site is completed (update-copy processing at the main site waits for completion of the update-copy processing at the remote site).

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

synchronous group

A group made up entirely of synchronous pair volumes.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

synchronous pair volume

A pair volume that has been created by specifying either data or never as the fence level. When data is written to the P-VOL, it is mirrored synchronously onto the S-VOL. When paired, no differences exist between the data on the P-VOL and the data on the S-VOL.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

system file

A system file stores information that will be needed if it becomes necessary to recover the system in the event of an error. System file is a generic term that includes the following files:

• System log files
• Synchronization point dump files
• Status files

system generator

The system generator is a facility that enables a HiRDB system to be constructed interactively by selecting displayed values.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

system log file

A system log file stores a history of database updates; it is also called a journal file. Historical information on database updates is called the system log (or system log information), and it is used by HiRDB to recover the database in the event of abnormal termination of either HiRDB or a UAP. The system log is also used as input information for recovery of the database.

system RDAREA

System RDAREA is a generic term for the following types of RDAREAs:

• Master directory RDAREA
• Data directory RDAREA
• Data dictionary RDAREA

system switchover facility

When a standby server machine is provided separately from the server machine that is processing jobs, job processing can be switched automatically by HiRDB to the standby server machine in the event the running server machine fails. This is called the system switchover facility.

table reorganization

Over time, additions and updates of data in a table tend to fragment the arrangement of rows, giving rise to unusable free space. Table reorganization reorganizes table data in a user RDAREA or LOB data in a LOB RDAREA to eliminate unusable free space. The database reorganization utility (pdrorg command) is used to reorganize a table.

tape device access facility

A facility that enables access to files on DAT, DLT, and LTO. The tape device access facility is used for the following types of files:

• Input data files (input data files specified in the source statement of the pdload command)
• Unload data files (unload data files specified in the unload statement of the pdrorg command)
• Unload data files (LOB data unload data files specified in the lobunld statement of the pdrorg command)
• Backup files (backup files specified with the -b option of the pdcopy or pdrstr command)

This term has no practical application for UNIX users, because it is related to a facility that cannot be used with a UNIX version of HiRDB.

transaction

A transaction is a unit of logical work; for example, a series of database operations. The transaction is also the basic unit for recovery and locking.

transaction execution site

In the log-only synchronous method, the site that accepts transactions.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

trigger

By defining a trigger, you can have SQL statements automatically execute when an operation (updating, insertion, deletion) is performed on a particular table. When a particular table is updated, a trigger based on the associated event allows you to automatically perform operations such as updating another table as well.

unbalanced index split

Unlike normal page split, the unbalanced index split method splits the data in an index page into two unequal parts, rather than into two equal parts. Index storage efficiency is improved by using this method when an ascending or descending middle key is to be added.

uniqueness constraint

The uniqueness constraint is a limitation that prohibits a data value from appearing more than once in a column (all data values in the column must be unique).

unit

A unit refers to the HiRDB operation environment within a single server machine.

unit controller

A unit controller is a system that controls and monitors server execution in a unit and that controls communication between units.

unload log file

A file created by unloading a system log file (system log information) with the pdlogunld command is called an unload log file.

unload statistics log file

A file created by unloading the contents of a statistics log file.

unused page

A page that is not being used.

unused segment

A segment that is not being used. Such a segment can be used by any table (or index) in the RDAREA.

updatable back-end server

A back-end server on which shared tables and shared indexes in shared RDAREAs can be updated.

updatable online reorganization

Refers to functionality that allows databases to be accessed and updated during database reorganization. Processes that access or update a database perform the operation on a replica database. To perform updatable online reorganization, you must install HiRDB Staticizer Option, and you must also specify related operands in the HiRDB system definitions.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

update buffer

When data is updated, the data is first updated into a global buffer, before being updated in the database. The buffer that stores this data before the database is updated is called an update buffer.

update copy

An operation in which, when updating of data on a primary volume occurs, the updated data is also mirrored (copied) onto a secondary volume. Update copying maintains integrity between the primary volume and the secondary volume.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

used page

A page in which a table or an index is stored. A used page that is completely filled with data such that no more can be added is called a full page, and a used page from which data has been deleted so that it no longer contains data is called a used free page.

used segment

A segment in which table or index data is stored. A used segment that is completely filled with data such that no more can be added is called a full segment, and a used segment from which data has been deleted so that only free pages remain (used free pages or unused pages) is called a used free segment.

user-defined type

A data type that can be defined by the user is called a user-defined type. Abstract data types are user-defined types.

user LOB RDAREA

A user LOB RDAREA stores large variable object data, such as documents, images, and sounds. The following types of data must be stored in a user LOB RDAREA:

• Column with the BLOB type specification (BLOB column)
• Attribute with BLOB specification in an abstract data type
• Plug-in index

user mapping

Information defining the user ID and password required to log into a foreign server. This information is defined for each foreign server and is used to establish connection.

user RDAREA

A user RDAREA stores tables and indexes created by a user.

user server hot standby

A facility that starts server processes on a standby HiRDB beforehand, so that startup processing of server processes is not performed when the system is switched over. The system switchover time is reduced by the amount of time required to perform startup processing of server processes.

In addition to user server hot standby, the rapid system switchover facility is available for reducing system switchover time. The rapid system switchover facility can reduce the system switchover time more than user server hot standby can (the rapid system switchover facility includes the user server hot standby functionality).

utility special unit

A utility special unit is a server machine in which only input/output devices to be used for executing a utility are set up. A utility special unit can be set up only in a HiRDB/Parallel Server. The following utilities can use a utility special unit:

• Database load utility
• Database reorganization utility
• Dictionary import/export utility
• Database copy utility
• Database recovery utility

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

value (instance)

The term value refers to a specific value of an abstract data type.

view table

A newly defined virtual table in which specific rows and columns are selected from a table that actually exists (called a base table).

If you are using HiRDB External Data Access, you can create a view table from foreign tables.

volume attribute

There are three types of volumes, primary volumes (P-VOL), secondary volumes (S-VOL), and simplex volumes (SMPL). The volume attribute refers to the attribute indicating one of these three volume types.

This term has no practical application for Windows users, because it is related to a facility that cannot be used with a Windows version of HiRDB.

work table file

A file that stores temporary information needed in order to execute an SQL statement is called a work table file.