[
https://issues.apache.org/jira/browse/IGNITE-17220?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17565056#comment-17565056
]
Alexander Belyak commented on IGNITE-17220:
-------------------------------------------
I ran out test on Ignite3 master, java11openjdk in docker, centos and get the
following results:
|3 node|all ops|1|13|
| | |2|13|
| |insert|1|234|
| | |2|173|
|1 node|all ops|1|8|
| | |2|8|
| |insert|1|202|
| | |2|181|
Used 4 servers (for each server node and an additional one for the benchbase
client): 2x Xeon E5-2609v4 16vCPU, 96GB, 2.4Tb SSD RAID0.
Start 3 nodes with the default config (only with all 3 nodes ip specified in
config), activate cluster with --meta-storage-node=<secondOneFor3NodeCluster>
and do all 4 benchbase tests (via the first node in jdbc string). Then I ran
clear single node cluster and repeat tests for it.
All logs and JFR in attachment.
> YCSB benchmark run for ignite2 vs ignite3
> -----------------------------------------
>
> Key: IGNITE-17220
> URL: https://issues.apache.org/jira/browse/IGNITE-17220
> Project: Ignite
> Issue Type: Task
> Reporter: Kirill Gusakov
> Assignee: Alexander Belyak
> Priority: Major
> Labels: ignite-3
>
> For further investigation of ignite3 performance issues, we need to run the
> following benchmarks to compare ignite2 vs ignite3 performance:
> * Usual ycsb benchmark with mixed load patterns
> * Insert-only ycsb benchmark
> For ignite2 and ignite3 in the following configurations:
> * 3 ignite nodes setup (so, table must have 1 partition and 3 replicas)
> * 1 ignite node setup (so, table must have 1 partitoin and 1 replica)
> Also, please provide:
> * Hardware configuration of the environment, where benchmark was executed
> * JFRs for every node in every run.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
