OpenTP1 Version 7 Programming Guide

[Contents][Index][Back][Next]

4.3.2 Outline of internode shared tables

This subsection outlines internode shared tables.

Organization of this subsection
(1) Environment for access to internode shared tables
(2) Internode shared table structure

(1) Environment for access to internode shared tables

Internode shared tables reside in shared memory at each node. There is no file equivalent to the entity of each table. Therefore, internode shared tables can be accessed from a UAP in online environment only. They cannot be accessed in offline environment.

When the IST service is used across multiple nodes, the time must match among these nodes. If not, data updates at one node may not be reflected at another node.

The figure below shows the processing flow when the IST service updates an internode shared table record (a record contained within an internode shared table) on multiple nodes.

Figure 4-20 Updating an internode shared table record

[Figure]

  1. The IST service creates record update data for updating an internode shared table record (record number 1) of the internode shared table A at node A.
  2. It acquires the current time (machine time, in microseconds) and confers it as a time stamp on the record update data.
  3. The IST service compares the time stamp set in the relevant internode shared table record in shared memory at node A and the time stamp given to the record update data.
    If the record update data is more recent, the IST service updates the internode shared table record in shared memory. If the record update data is older, the IST service does not update the internode shared table record in shared memory. Even when the IST service does not update the internode shared table record, the function dc_ist_write() returns normally.
  4. When the IST service has updated the internode shared table record in shared memory, it notifies the IST service at node B that it has updated an internode shared table record at node A. At this time, it also reports the internode shared table record and the time stamp given to the internode shared table record.
  5. The IST service at node B which received the updated internode shared table record compares the time stamp set in the relevant internode shared table record within the node and the time stamp of the internode shared table record that it received.
  6. Only if the IST service determines as the result of step 5 that the time stamp of the internode shared table record that was received is more recent does it update the relevant internode shared table record at node B to the information provided in the internode shared table record that was received.

As explained above, the IST service determines whether to update an internode shared table record or leaves it as it is based on the time stamp. In the following cases, the latest update data may not be reflected in the internode shared table record.

(2) Internode shared table structure

References and updates of an internode shared table from a UAP are done in units of records. An internode shared table consists of multiple records. A UAP process can access one record or access two or more records collectively.