[
https://issues.apache.org/jira/browse/CASSANDRA-18831?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18056327#comment-18056327
]
Dmitry Konstantinov edited comment on CASSANDRA-18831 at 2/3/26 8:54 PM:
-------------------------------------------------------------------------
An extra observation for ZGC: it feels much better if we reduce memtable max
heap size in conf/cassandra.yaml
{code:java}
memtable_heap_space: 2048MiB # by default Cassandra uses 1/4 of heap size, in
our case it is 7887MiB {code}
with this change I do not observe allocation stalls anymore even for 200
threads and have a much higher throughput (getting close to G1) with a good
enough latency:
||ZGC, JDK21, 100 client threads||ZGC, JDK21, 200 client threads||
|{{Op rate : 163,608 op/s [insert: 163,608 op/s]}}
{{Partition rate : 163,608 pk/s [insert: 163,608 pk/s]}}
{{Row rate : 1,636,084 row/s [insert: 1,636,084 row/s]}}
{{Latency mean : 0.6 ms [insert: 0.6 ms]}}
{{Latency median : 0.5 ms [insert: 0.5 ms]}}
{{Latency 95th percentile : 0.9 ms [insert: 0.9 ms]}}
{{Latency 99th percentile : 1.2 ms [insert: 1.2 ms]}}
{{Latency 99.9th percentile : 11.6 ms [insert: 11.6 ms]}}
{{Latency max : 35.9 ms [insert: 35.9 ms]}}
{{Total partitions : 15,000,000 [insert: 15,000,000]}}
{{Total errors : 0 [insert: 0]}}
{{Total GC count : 178}}
{{Total GC memory : 220.252 GiB}}
{{Total GC time : 51.9 seconds}}
{{Avg GC time : 291.5 ms}}
{{StdDev GC time : 571.7 ms}}
{{Total operation time : 00:01:31}}|{{Op rate : 193,924
op/s [insert: 193,924 op/s]}}
{{Partition rate : 193,924 pk/s [insert: 193,924 pk/s]}}
{{Row rate : 1,939,237 row/s [insert: 1,939,237 row/s]}}
{{Latency mean : 1.0 ms [insert: 1.0 ms]}}
{{Latency median : 0.8 ms [insert: 0.8 ms]}}
{{Latency 95th percentile : 1.8 ms [insert: 1.8 ms]}}
{{Latency 99th percentile : 3.7 ms [insert: 3.7 ms]}}
{{Latency 99.9th percentile : 20.5 ms [insert: 20.5 ms]}}
{{Latency max : 76.1 ms [insert: 76.1 ms]}}
{{Total partitions : 15,000,000 [insert: 15,000,000]}}
{{Total errors : 0 [insert: 0]}}
{{Total GC count : 224}}
{{Total GC memory : 80.476 GiB}}
{{Total GC time : 71.0 seconds}}
{{Avg GC time : 317.0 ms}}
{{StdDev GC time : 646.6 ms}}
{{Total operation time : 00:01:17}}|
was (Author: dnk):
An extra observation for ZGC: it feels much better if we reduce memtable max
heap size in conf/cassandra.yaml
{code:java}
memtable_heap_space: 2048MiB # by default Cassandra uses 1/4 of heap size, in
our case it is 7887MiB {code}
with this change I do not observe allocation stalls anymore even for 200
threads and have a higher throughput with a good enough latency:
||ZGC, JDK21, 100 client threads||ZGC, JDK21, 200 client threads||
|{{Op rate : 163,608 op/s [insert: 163,608 op/s]}}
{{Partition rate : 163,608 pk/s [insert: 163,608 pk/s]}}
{{Row rate : 1,636,084 row/s [insert: 1,636,084 row/s]}}
{{Latency mean : 0.6 ms [insert: 0.6 ms]}}
{{Latency median : 0.5 ms [insert: 0.5 ms]}}
{{Latency 95th percentile : 0.9 ms [insert: 0.9 ms]}}
{{Latency 99th percentile : 1.2 ms [insert: 1.2 ms]}}
{{Latency 99.9th percentile : 11.6 ms [insert: 11.6 ms]}}
{{Latency max : 35.9 ms [insert: 35.9 ms]}}
{{Total partitions : 15,000,000 [insert: 15,000,000]}}
{{Total errors : 0 [insert: 0]}}
{{Total GC count : 178}}
{{Total GC memory : 220.252 GiB}}
{{Total GC time : 51.9 seconds}}
{{Avg GC time : 291.5 ms}}
{{StdDev GC time : 571.7 ms}}
{{Total operation time : 00:01:31}}|{{Op rate : 193,924
op/s [insert: 193,924 op/s]}}
{{Partition rate : 193,924 pk/s [insert: 193,924 pk/s]}}
{{Row rate : 1,939,237 row/s [insert: 1,939,237 row/s]}}
{{Latency mean : 1.0 ms [insert: 1.0 ms]}}
{{Latency median : 0.8 ms [insert: 0.8 ms]}}
{{Latency 95th percentile : 1.8 ms [insert: 1.8 ms]}}
{{Latency 99th percentile : 3.7 ms [insert: 3.7 ms]}}
{{Latency 99.9th percentile : 20.5 ms [insert: 20.5 ms]}}
{{Latency max : 76.1 ms [insert: 76.1 ms]}}
{{Total partitions : 15,000,000 [insert: 15,000,000]}}
{{Total errors : 0 [insert: 0]}}
{{Total GC count : 224}}
{{Total GC memory : 80.476 GiB}}
{{Total GC time : 71.0 seconds}}
{{Avg GC time : 317.0 ms}}
{{StdDev GC time : 646.6 ms}}
{{Total operation time : 00:01:17}}|
> JDK21 support
> -------------
>
> Key: CASSANDRA-18831
> URL: https://issues.apache.org/jira/browse/CASSANDRA-18831
> Project: Apache Cassandra
> Issue Type: Task
> Components: Build
> Reporter: Achilles Benetopoulos
> Assignee: Josh McKenzie
> Priority: Normal
> Labels: pull-request-available
> Fix For: 5.1
>
> Attachments: batch_profile_seq.yaml,
> image-2026-02-03-19-05-41-619.png, image-2026-02-03-19-16-37-640.png,
> jdk21-patch, jdk_gc.html, jdk_gc_g1_zgc_shen.html
>
> Time Spent: 9h
> Remaining Estimate: 0h
>
> This patch builds on the work in CASSANDRA-16895 that added JDK17 to the list
> of supported Java versions, and extends that work to enable building and
> running Cassandra under JDK21.
> The following commits comprise the changes included in the attached patch:
> -
> [https://github.com/apache/cassandra/commit/b15d4d6980e787ab5f3405ca8cb17a9c92a4aa47]
> -
> [https://github.com/apache/cassandra/commit/0c5df38dafe58bfec8924e81507bb604e1543897]
> -
> [https://github.com/apache/cassandra/commit/6506b7279d98eed4b2b65b71e0c6f41eb78c6913]
> -
> [https://github.com/apache/cassandra/commit/564cbd534c5a975cda0c629c14c68c8745b41451]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
