Ansicht
Dokumentation

RILM_STOR_TEST_PET - ILM Store: Performance Evaluation Tool

CPI1466 during Backup TXBHW - Original Tax Base Amount in Local Currency
This documentation is copyright by SAP AG.

Purpose

You use the Performance Evaluation Tool (PET) report to evaluate the performance of the ILM Store operations.

The ILM Store is a new storage solution for data retention. The ILM Store supports different types of storage, for example, SAP HANA, Sybase IQ, file systems, and Hadoop. The performance of the ILM Store is influenced by the type of database you use and by settings that you make in Customizing.

Note: If you want to execute the report as a background job and export the results into a file, you must select the target application server on which to run the report. File management (that is, file creation and deletion) is the ILM Store user’s responsibility.

Prerequisites

The ILM Store and underlying storage media are available in your system and you have made the relevant settings in Customizing. For more information about the configuration of the ILM Store, see the Installation and Configuration Guide for the ILM Store.

Features

To allow performance evaluation for the target storage media, processing automatically switches to synchronous mode, which means that data is directly written into the media. The setup returns to its original state when the performance evaluation is complete.

The PET exports measured results into a file. This feature allows further analysis of collected statistical data and enables the collection of data through background processing.

The PET carries out the performance evaluation by default for the ILM Store’s test operational origin (archeb). The target storage media depends on the settings that you have made for the origin in Customizing.

The PET evaluates the performance of the ILM Store operations in the following order:

TRANSFER - Adds resource data to the internal memory
INSERT - Writes resource to storage
READ - Reads resource from storage
READ CHUNK - Reads part of a resource from storage
PPATCH - Creates or deletes property
PFIND - Reads property
DELETE - Deletes resource from storage

The operations’ performance can change depending on the depth of the target object (resource or collection) and its distance from the root collection. To differentiate the performance based on the depth of the target object, the PET measures two extremes:

The performance of an operation on a resource in the root collection (at the beginning of a collection tree)

The performance of an operation on a resource in a leaf collection (at the end of a collection tree)

Note: To simulate the complex branching of collections within the ILM Store, the PET uses a collection tree. The collection tree is a binary tree that branches to a maximum depth. The nodes of the tree represent collections and each collection branches into two new collections (binary branching). The first collection of the tree is the root collection. The collections that reach the maximum depth, named leaf collections, do not branch further.

The maximum depth of the collection tree is eight levels. This number is set based on the known collection path from which the collection tree starts to branch and the maximum collection depth defined by the ILM Store.

The following collection tree is created during the execution of the PET report:

//pet///

The placeholders are as follows:

- The default collection Uniform Resource Identifier (URI) used for testing. It is defined as a constant GC_TEST_URI_PREFIX in the interface IF_ILM_STOR_TYPES.

- The operational origin under which the PET report is executed.

- The connection ID defined in operational Customizing: Origin ; namespace DB; property DBCON.TILM_STOR_BLOB

- This placeholder represents all collections on the collection tree.

The system performs a cleanup when the program execution ends and deletes the collection tree //pet with all of its child collections.

Selection

To run the report, you make the following settings:

Origin – Enter the operational origin.

Binary Large Objects (BLOBs) Size- Choose the data sizes upon which the operations are performed.

Iterations – Choose the number of iterations for each store operation; the more iterations the more accurate the resulting statistics. The maximum value is 999.

File Name – Enter the name of the file on an application server to which the performance evaluation results are exported. If you do not specify a name, the results are displayed on the screen.

Note: You can view the results stored in the file using transaction AL11. Make sure that you are on the correct application server.

Overwrite - Determines whether the file is overwritten by new test results. If you do not make this selection, the system appends the results to the existing file.

Output

If you specify a name in the File Namefield, the results are either stored in the chosen file or displayed on the screen in a table. The results are exported to the file in comma separated values (CSV) format with the delimiting character ';'. The data is structured as follows:

;;;;;;;;;;;