[ 
https://issues.apache.org/jira/browse/IGNITE-20882?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17800470#comment-17800470
 ] 

Andrey Novikov edited comment on IGNITE-20882 at 1/24/24 8:28 AM:
------------------------------------------------------------------

 
Single-node embedded (CriteriaSingleNodeBenchmark):
||Benchmark||Score(us/op)||
|criteriaGet|1780,171 ± 35,306|
|kvGet|121,114 ±  1,726|
|sqlGet|1703,928 ± 34,945|

Two nodes, thin client(CriteriaThinClientBenchmark):
||Benchmark||Score(us/op)||
|criteriaGet|1780,171 ± 35,306|
|kvGet|251,105 ± 8,618|
|sqlGet|2694,740 ± 72,396|
|criteriaGetNonNullTxDisablesPartitionAwareness|2935,503 ± 192,479|
|kvGetNonNullTxDisablesPartitionAwareness|352,037 ± 7,819|
|sqlGetNonNullTxDisablesPartitionAwareness|2814,066 ± 88,770|
|sqlIterate|1603991,380 ± 12153,762|
|criteriaIterate|1501173,640 ± 31600,562|

 


was (Author: anovikov):
 
Single-node embedded:
||Benchmark||Score(us/op)||
|CriteriaSingleNodeBenchmark.criteriaGet|1780,171 ± 35,306|
|CriteriaSingleNodeBenchmark.kvGet|121,114 ±  1,726|
|CriteriaSingleNodeBenchmark.sqlGet|1703,928 ± 34,945|

 

 

> Append performance tests
> ------------------------
>
>                 Key: IGNITE-20882
>                 URL: https://issues.apache.org/jira/browse/IGNITE-20882
>             Project: Ignite
>          Issue Type: Improvement
>          Components: sql
>            Reporter: Andrey Novikov
>            Assignee: Andrey Novikov
>            Priority: Major
>              Labels: ignite-3
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> # Compare single-key criteria vs single-key SQL vs table.get(PK)
>  * Expect to have similar performance
>  * Check single-node embedded, distributed with no partition awareness, 
> distributed with partition awareness
>  * NOTE: future improvement for criteria query – lookup affinity node if PK 
> or colocation key is in the criteria.
>  # Compare criteria vs SQL when affinity key is specified (unicast when 
> partition pruning is added)
>  # Compare criteria vs SQL for broadcast queries – latency and throughput
>  # Compare criteria vs SQL – time it takes to iterate over 10kk entries



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to