phoenix-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yonatan Gottesman (JIRA)" <>
Subject [jira] [Commented] (PHOENIX-5082) Observations about the Phoenix-Omid integration
Date Fri, 28 Dec 2018 13:21:00 GMT


Yonatan Gottesman commented on PHOENIX-5082:

Hi [~lhofhansl], I started playing around with your test.

The omid client has 2 modes to update the shadow cells (and then remove entry from commit
table): SYNC and ASYNC.

The default is SYNC witch means that the same thread that does the put is going to write the
shadow cell and delete commit entry -  *after the commit is done.*

There is an ASYNC mode that delegated these actions to a background thread (might cause other
gets to be slower because shadow cells are not updated right away witch means another hop
to commit table).

With async mode 130K rows takes 27 sec:
0: jdbc:phoenix:> upsert into test select rand()*10000000, rand()*10000000, rand()*10000000,
rand(), rand(), rand()*1000000 from test;
131,072 rows affected (27.632 seconds)
I want you to give it a try:

create a file named hbase-omid-client-config.yml and put it in your classpath.

In that file have this line:
postCommitMode: !!org.apache.omid.tso.client.OmidClientConfiguration$PostCommitMode ASYNC
ill keep looking why this still takes 5 times more than without transactions


> Observations about the Phoenix-Omid integration
> -----------------------------------------------
>                 Key: PHOENIX-5082
>                 URL:
>             Project: Phoenix
>          Issue Type: Task
>            Reporter: Lars Hofhansl
>            Assignee: Yonatan Gottesman
>            Priority: Major
>         Attachments: CommitTS-Sample.png, Omid-profile.png, Omid-sample.png
> Filing this issue to track my observations and perhaps some fixes.
> [~ohads]
> [~jamestaylor]
> [~yonigo]

This message was sent by Atlassian JIRA

View raw message