[jira] [Commented] (ACCUMULO-3813) single row look-ups are slow

["Dylan Hutchison (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ACCUMULO-3813?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14544076#comment-14544076 ] 

Dylan Hutchison commented on ACCUMULO-3813:
-------------------------------------------

I think this is the same issue I found for ACCUMULO-3710, only in my case the tserver ran out of memory.  Accumulo doesn't handle large numbers of small, disjoint ranges well.  I bet there's room for improvement on both the client and tablet server.

> single row look-ups are slow
> ----------------------------
>
>                 Key: ACCUMULO-3813
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-3813
>             Project: Accumulo
>          Issue Type: Improvement
>          Components: client, tserver
>            Reporter: Eric Newton
>
> Emilio Lahr-Vivaz writes on the user mailing list:
> {quote}
> I've found that scanning lots of non-sequential single-row ranges is pretty slow in accumulo. Your best approach is probably to create an index table on whatever you are originally trying to query (assuming those 10000 ids came from some other query).
> {quote}
> Specifically, the use case is fetching many single items, all of which are present (so, bloom filters aren't going to help).
> Since this is one of the use cases Accumulo was designed to handle, look into actual performance and figure out if there are any obvious bottlenecks.



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