Red Hat Directory Server 7.1 for HP-UX11i v2 Integrity System Performance Tuning and Sizing Guideline

Red Hat Directory Server 7.1 Performance Tuning and Sizing Guidelines
• RHDS parameter settings: dbcachesize and cachememsize are big enough to cache all the entries, and
nsslapd-threadnumber is set to 6.
Server DirMark
Search
performance
Performance Increases as #of
CPUs increases (compare to 2
CPU only)
Montecito-based test configuration 2-CPUs @1.6GHz 5135.46
Montecito-based test configuration 4-CPUs @1.6GHz 8337.53 62%
Montecito-based test configuration 6-CPUs @1.6GHz 9999.68 95%
Montecito-based test configuration 8-CPUs @1.6GHz 13106.38 155%
Table 1: RHDS 7.1 performance with different number of CPUs on a Montecito-based test configuration
@1.6GHz/CPU
From Table 1, we can see that as the number of CPUs increased, the performance is getting better.
Data Collection 2: (Different nsslapd-threadnumber)
This set of data is collected to help us understand how number of worker threads in ns-slapd process affects the
performance.
• Server: Montecito-based test configuration with 8 CPUs @1.6 GHz each
• #of entries: 100K entries
• RHDS parameter setting: dbcache and entry cache are big enough to hold all the pages, entries.
nsslapd-threadnumber changes as indicated in Table 2.
nsslapd-threadnumber DirMark Search performance
2 5340.03
4 9789.96
6 13209.48
8 12874.27
12 12069.33
16 10613.60
24 9634.56
32 8656.48
64 5183.54
96 5217.02
128 5384.33
Table 2: RHDS7.1 performance with different settings of nsslapd-threadnumber
From Table 2, we can see that as number of thread increases, the search throughput gets worse. Please note when
nsslapd-threadnumber drops from the default of 32 to 6, the exact search throughput increased 53%.
Data Collection 3: (Different nsslapd-dbcachesize)
This set of data is collected to help us understand how dbcachesize affects the performance:
Page 12