Capacity planning

Edit online

Before you install IBM® Verify Directory on a system, you must decide the hardware to use for setting up the existing capacity and scaling up when required.

The hardware resources that you must consider before you install and deploy IBM Verify Directory are:

  • Hard disk
  • Memory
  • Processors

IBM Verify Directory performance on various hardware configurations might vary. It is important to understand the hardware configurations that can provide the system resources to the directory server for best performance.

Several tuning measures were taken to find the best settings under which the directory server provides the best results. Tuning factors tested were:

IBM Verify Directory tuning
LDAP entry cache
DB2® tuning
DB2 buffer pools
  • LDAP buffer pool
  • IBMDEFAULT buffer pool
Optimization and organization (reorgchk and reorg)
DB2 configuration parameters
Backing up and restoring the database (backup and restore)
Splitting of database

The capacity planning information includes results that are observed from uploading data with the bulkload utility and running specific benchmarks on various AIX® systems.

The results that are provided are obtained through the following methods:

  • Two types of Lightweight Directory Interchange Format (LDIF) files were used. The first type of LDIF file contains small entries with a flat tree structure. The other LDIF file contains larger entry sizes and a deeper tree structure.
  • LDIF files with 100,000 and 1 million entries were created. The first type of LDIF file was used only for search operations. The second type was used for both searches and updates.

In each case, a directory server instance is created on the system. The LDIF file is loaded into the database and the recording were made. Operating system information and information that is related to the directory server instance is collected at intervals throughout the load and the performance test run.

Note: The statistics reported here are specific to the particular hardware setups used and were generated in a lab environment. These results might not be reproducible in other environments. The results reported here must be used only as guidelines.

The directory server performance was monitored by running various workloads on different hardware configurations. These workloads were run on a number of AIX systems.