Hi all, I'm on my way to migrating from a proprietary search technology (Endeca) to SolrCloud 4.6.1.
I'm running 4 servers with 8 cores (2 threads per core with hyperthreading) and 64 Gb of RAM per server. Each server has 4 shards and each shard is running in a separate Tomcat instance with a 4 Gb heap. I'm using 5 external Zookeeper instances (1 per Solr server and 1 on our ingestion api server). Our document count is currently very small at about 300k documents total. I've gotten our 95th percentile query time down to ~30 msec by tuning the cache sizes in solrconfig.xml for each shard. The sizes are: filterCache: size=16384, initialSize=4096, autowarmCount=0 queryResultCache: size=512, initialSize=512, autowarmCount=0 documentCache: size=2048, initialSize=2048, autowarmCount=0 fieldValueCache: size=512, initialSize=512, autowarmCount=128, showItems=32 I'm trying to get the query time down to ~15 msec. Anyone have any tuning recommendations? -- joel cohen, senior system engineer e joel.co...@bluefly.com p 212.944.8000 x276 bluefly, inc. 42 w. 39th st. new york, ny 10018 www.bluefly.com <http://www.bluefly.com/?referer=autosig> | *fly since 2013...*
