cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chris Burroughs (JIRA)" <j...@apache.org>
Subject [jira] Commented: (CASSANDRA-1966) Option to control how many items are read on cache load
Date Wed, 12 Jan 2011 01:40:46 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-1966?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12980493#action_12980493
] 

Chris Burroughs commented on CASSANDRA-1966:
--------------------------------------------

I don't understand your comment about 10% of the cache being basically useless.  Wouldn't
this depend  on the size of the cache, time to read rows, distribution of requests, etc.?

At least in my case I have small rows and and  a row cache sized to 400,000 items.  With a
cold cache a large (most?) numbers of pending tasks in the ROW-READ-STAGE are dropped (this
lasts for 10s of minutes).  While the node has obviously joined the cluster at this point
I would not consider it functional.  Once the cache reaches around 30,000 items with a RecentHitRate
around 30%  ROW-READ-STAGE tasks stop being dropped.  Once the row cache fills at 400,000
the recent RecentHitRate hovers close to 40% (as expected diminished returns).

30,000 is 7.5% of the cache.

> Option to control how many items are read on cache load
> -------------------------------------------------------
>
>                 Key: CASSANDRA-1966
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1966
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: Chris Burroughs
>
> CASSANDRA-1417 added an option to save the key and/or row cache keys which is cool. 
However, for a row large cache it can take a long time to read all of the rows.  For example
I have a 400,000 item row cache, and loading that on restart takes a little under an hour.
> In addition to configuring the size of the row cache, and how often it should be saved
to disk, I propose an option to control how many items are loaded on startup (or alternately
only saving n items out of the full row cache to begin with).

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message