cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ariel Weisberg (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (CASSANDRA-7217) Native transport performance (with cassandra-stress) drops precipitously past around 1000 threads
Date Sat, 14 Nov 2015 19:50:11 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-7217?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Ariel Weisberg updated CASSANDRA-7217:
--------------------------------------
    Attachment: FakeQuerySystem.java

To test stress and threading in general I mocked out interactions between stress and the client
library. There is no performance regression up to 4000 threads if you remove the server and
client library from the picture.

Attached is what I used to fake the queries. It's a thread pulling queries off a delay queue
and the delay is set to be a uniform distribution between some minimum and maximum latency.
I tried 3-9 milliseconds with a server side throughput of 100k. The threads issuing the queries
are woken up via {{java.util.concurrent.FutureTask}}.

I'll mock out the server in the client library next.

> Native transport performance (with cassandra-stress) drops precipitously past around
1000 threads
> -------------------------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-7217
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-7217
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Benedict
>            Assignee: Ariel Weisberg
>              Labels: performance, stress, triaged
>             Fix For: 3.1
>
>         Attachments: 2000-threads.svg, 500-threads.svg, FakeQuerySystem.java
>
>
> This is obviously bad. Let's figure out why it's happening and put a stop to it.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message