3.2.1 Performance data
The data that is collected from the monitoring agent and indicates details of operation status is called performance data. The system administrator can check the collected performance data by using a Web browser.
Performance Management collects performance data in units called records. Items that can be monitored vary depending on PFM - Agent or PFM - RM.
There are two types of performance data:
- Real-time data
-
This type of performance data displays the current status of the monitored target.
Use this type to check the status of a constantly changing system and system errors in the window of PFM - Web Console.
- Historical data
-
This type of performance data displays the status of the monitored target from past to present.
Use this type mainly to analyze system trends. The monitoring agent database stores historical data in one of the following two formats depending on data characteristics:
-
Summary record
The system automatically calculates the values collected by the monitoring agent, summarizes the average and total values in minutes, and hourly, daily, weekly, monthly, and yearly units, and then the Store database stores this summary data.
The summary record monitors long-term system performance. These records are called a Product Interval record type (PI record type).
-
Non-summary record
The Store database stores the performance data in the same state as collected by the monitoring agent.
There are two record types: A Product Detail record type (PD record type) that collects the performance data displaying the status of the system at a certain point (for example, detailed information of a currently operating process), and a Product Log record type (PL record type) that collects logs and messages of systems and applications.
Record types other than above are available depending on the monitored target. For details, see the appropriate chapters that describe records in each PFM - Agent or PFM - RM manual.
-
- Organization of this subsection
(1) Records and fields
In Performance Management, the monitoring agent collects and manages operating information from the monitored targets in units called records and fields.
- Record
-
A record is a set of monitoring items categorized according to purpose. For example, the System Overview (PI) record is a set of monitoring items that provide an overview of the system, such as CPU usage and size available in the physical memory area. The monitoring agent collects this operating information in units of records. The records that can be collected vary depending on the agent program.
- Field
-
A field is an individual monitoring item that is part of a record. For example, in the System Overview (PI) record, CPU % and Page Faults/sec are fields.
The following figure shows the concepts of records and fields, using the System Overview (PI) record as an example.
For details about the records and fields that can be collected by each agent program, see the explanation of records in the manual of the relevant agent program.
(2) Data models
The Store database stores performance data in record units. Each record is divided into smaller units, called fields.
The generic term for the performance data collected by agents is called a data model. Data models are managed by version.
|
(3) Single-instance and multi-instance records
There are two types of performance data records collected by Performance Management depending on the monitoring items:
- Single-instance record:
-
A record that collects one record at a time is called a single-instance record.
Figure 3‒8: Example of a single-instance record - Multi-instance record:
-
A record that collects multiple records at a time is called a multi-instance record.
Figure 3‒9: Example of a multi-instance record - Point:
-
A group of records collected at the same time is called a data group. A data group of a single-instance record consists of one record. A data group of a multi-instance record consists of multiple records.