Suggestions on reviewing and improving and existing Directory implementation.
Review the Hardware Configuration
- Appropriate server and disk sizing: Enough CPUs, Disks, RAM.
- Adequate capacity planning for future growth: Estimate current growth and predict when the existing system will be over loaded.
- Options for increasing performance by adding hardware: Can the backplane/chassis accommodate additional CPUs, Disks, RAM.
Review the Network Configuration and Monitoring
Review the OS configuration
- Adequate bandwidth. Normally LDAP operations are small, however if large binary object are stored in the directory or there are many applications or concurrent connections bandwidth may cause a problem.
- Low latency. High latency throttles response time in synchronous replicated environments.
- If multiple NICs are configured then ensure both networks have been independently tested.
- Running processes on each server, in addition to eTrust Directory, these indicate if the box is not dedicated to DXserver. This may explain inconsistency in performance.
- Check the OS scheduler doesn't contain batch jobs that require large resources, database reporting for example.
- Disk striping/mirroring and RAID levels. If maybe possible to improve performance by changing these.
- Disk fragmentation, another performance killer.
Review the Ingres configuration
- Improvements in performance by adding locations, tuning the transaction log, tuning the DMF cache.
- If journaling is enabled for disaster recovery.
- If checkpoints are being taken regularly and how many are being kept both on disk and on tape.
Review the DXserver configuration
- Holes in connectivity: How is the DXserver knowledge shared?
- Failover: Are alternate DSAs defined?
- Security: Are the security flags min-auth and auth-levels consistent across the backbone?
- Performance Tuning: Is DXcache configured?
Review the DXserver Logs
- Alarms and Warning indicating problems in configuration or operation.
- The statistics logs over a 24 hour period. Graph each DSA log to show load profiles.
- The query logs can be used to separately graph the profile of searches and updates.
- The level of trace is not too high, choking performance.
Review the Disaster Recovery Plan and Operational Procedures
- A disaster recovery plan exists and includes contingency for the directory servers.
- Operational procedures include step-by-step instructions to perform: OS and Ingres backups/restores, DXserver management, Ingres management and tuning.
- The operational team responsible for monitoring and managing the directory are sufficiently trained.
- The DR plan and operational procedures are regularly tested.