[
https://issues.apache.org/jira/browse/CASSANDRA-749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12849317#action_12849317
]
Stu Hood commented on CASSANDRA-749:
------------------------------------
> it appears that they use nonlocal indexes, and it's so slow that they've
> "improved" it by automatically retrying timed-out queries
First, they're hosted on Bigtable, so of course they need timeouts. Second,
they have a section titled "Constraints Precede Performance", which shows why
you should constrain queries rather than allowing queries that will be slow.
> it's either ass because you have to do read base/update index/update index
> new/write base new
This applies to both local and distributed indexes. You will have to decide
between eager and lazy cleanups of the view, whether you are updating it
locally or remotely. The performance hit in the write path for the distributed
index is having to do blocking writes to W replicas in the view before starting
writes to the base.
> There is a race: if a reader sees an "empty" cell, and fills it in before the
> write completes, it will be
> marked as empty until the recency expires.
This was one of the points of that timeout parameter being set higher than a
write might take to complete: if a write was still in progress (view clock
doesn't match base clock), we wouldn't populate the view.
> Secondary indices for column families
> -------------------------------------
>
> Key: CASSANDRA-749
> URL: https://issues.apache.org/jira/browse/CASSANDRA-749
> Project: Cassandra
> Issue Type: New Feature
> Components: Core
> Reporter: Gary Dusbabek
> Assignee: Gary Dusbabek
> Priority: Minor
> Fix For: 0.8
>
> Attachments: 0001-simple-secondary-indices.patch,
> views-discussion-2.txt, views-discussion.txt
>
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
