Solr Nightly Benchmarks
Here a script runs every night to check out latest Solr trunk code and run benchmarks. The benchmark uses tweets as test data set, which is a good simulation of real time analytics use cases. You can find the test script at SOLR-10317.
In addition, there are lightweight benchmarks against Solr dependencies run and results are reported.
The goal is to 1) catch regressions 2) provide a performance baseline for capacity planning.
There are a couple of other Solr benchmarks that uses different data set and covers different use cases.
More details will come up.
JSON facet (Single Server)
latency.avg
throughput
CPU util
Term query (Single Server)
latency.avg
latency.p90
throughput
HDFS hsync (on 6-node CDH cluster)
latency.avg
latency.p90
throughput
S3 read (US-West-2 region)
latency for first byte
throughput
Single server configuration
Hardware
CPU: 24 cores Intel(R) Xeon(R) CPU E5-2630L v2 @ 2.40GHz
Memory: 128G 1333M DDR3
Disk: 12x 2TB disks. Type: ST2000NM0033-9ZM175 7200RPM. JBOD, No RAID.
Network: 10G
Solr
Heap: 4G
CDH cluster hardware
Nodes: 6 Solr nodes, co-living with 6 DN nodes. One extra node for HDFS NN, ZK, CM etc.
CPU: 48 cores Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz with hyper threading enabled
Memory: 96G 1333M DDR3
Disk: 12x 2TB disks. Type: ST2000NM0033-9ZM175 7200RPM. JBOD, No RAID.
Network: 10G