[ https://issues.apache.org/jira/browse/IGNITE-20812?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17829815#comment-17829815 ]
Iurii Gerzhedovich commented on IGNITE-20812: --------------------------------------------- [~zstan] results of check: sha: fd3d6a909b283c277c04f63ae, cpu: Intel Core i*-8000U ||*Benchmark*||*(clusterSize)*||*(fsync)*||*(partitionCount)*||*Mode*||*Cnt*||*Score*||*Error*||*Units*|| |InsertBenchmark.sqlInsert|1|false|1|avgt|20|0.318|± 0.058|us/op| |InsertBenchmark.sqlInsert|1|false|5|avgt|20|0.340|± 0.053|us/op| |InsertBenchmark.sqlInsert|1|false|25|avgt|20|0.364|± 0.056|us/op| |InsertBenchmark.sqlInsert (modified)|1|false|1|avgt|20|0.633|± 0.167|us/op| |InsertBenchmark.sqlInsert (modified)|1|false|5|avgt|20|0.660|± 0.147|us/op| |InsertBenchmark.sqlInsert (modified)|1|false|25|avgt|20|0.672|± 0.147|us/op| |SelectBenchmark.sqlGet|1|false|1|avgt|20|45.882|± 7.328|us/op| |SelectBenchmark.sqlGet|1|false|5|avgt|20|40.871|± 1.794|us/op| |SelectBenchmark.sqlGet|1|false|25|avgt|20|41.499|± 1.779|us/op| InsertBenchmark.sqlInsert (modified) - mean with removed {{HEP_TO_SIMPLE_KEY_VALUE_OPERATION}} rule, i.e. no single kv api is used here. > Sql. Performance of a queries affected by number of partitions > -------------------------------------------------------------- > > Key: IGNITE-20812 > URL: https://issues.apache.org/jira/browse/IGNITE-20812 > Project: Ignite > Issue Type: Improvement > Components: sql > Reporter: Konstantin Orlov > Priority: Major > Labels: ignite-3 > > Performance of INSERT and SELECT by primary key queries is affected by number > of partitions in the table. Most probably, the reason is a lack of partition > pruning in sql engine. > Let's investigate and fix the problem. > Here is a result of local (on my laptop; 2,6 GHz 6-Core Intel Core i7 32GB > RAM) benchmark: > ||Benchmark||(clusterSize)||(fsync)||(partitionCount)||Mode||Cnt||Score||Error||Units|| > |InsertBenchmark.sqlInsert|1|false|1|avgt|20|0.643|± 0.069|ms/op| > |InsertBenchmark.sqlInsert|1|false|5|avgt|20|0.756|± 0.049|ms/op| > |InsertBenchmark.sqlInsert|1|false|25|avgt|20|1.114|± 0.093|ms/op| > |SelectBenchmark.sqlGet|1|false|1|avgt|20|203.432|± 16.617|us/op| > |SelectBenchmark.sqlGet|1|false|5|avgt|20|320.383|± 22.086|us/op| > |SelectBenchmark.sqlGet|1|false|25|avgt|20|794.232|± 49.473|us/op| -- This message was sent by Atlassian Jira (v8.20.10#820010)