cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Stu Hood (JIRA)" <>
Subject [jira] Commented: (CASSANDRA-389) SSTable Versioning
Date Thu, 14 Jan 2010 07:06:54 GMT


Stu Hood commented on CASSANDRA-389:

I'm fine with storing the version and other metadata in the file (like 521 begins to do),
as long as the different implementations of sstables can agree on the location of this information.
For instance, the implementation in 674 stores SSTable metadata at the beginning of every
block so that the sstable is easily splittable for use by the Streaming API or by Hadoop at
some point in the future.

Perhaps version information is the one thing that should be stored outside the file, because
it can define how every other piece of metadata is stored?

> SSTable Versioning
> ------------------
>                 Key: CASSANDRA-389
>                 URL:
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: Chris Goffinet
>            Assignee: Stu Hood
>            Priority: Minor
>             Fix For: 0.9
>         Attachments: 389-v3.patch, 389-v4.patch, 389-v5-1-rebase-v4-for-trunk.diff, 389-v5-2-add-keyspace-to-descriptor.diff,
389-v5-3-use-descriptor-for-streaming.diff, 389-v5-4-validate-parameters-for-descriptor.diff,
389-v5-5-special-case-to-preserve-legacy.diff, CASSANDRA-389.diff, CASSANDRA-389.diff
> As we continue to make changes to the on-disk format of SSTables, I propose we start
versioning. The easiest way without breaking backwards compatibility is to store the version
in the filename. This would allow us to figure out the version without looking at the SSTable
data. After speaking to Jonathan here is the proposed example:

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

View raw message