[
https://issues.apache.org/jira/browse/TS-745?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13024839#comment-13024839
]
John Plevyak commented on TS-745:
---------------------------------
Overall several comments:
1) Once we get it working we need to reorganize the patch so that we avoid code
duplication.
2) The limitation of a single SSD is probably overly restrictive. SSDs are
relatively inexpensive now and it is very likely that folks will want to have 1
per disk or better yet some combination of SSD and SATA.
3) The code uses xmalloc to store docs and creates another CacheVC to do the
write in the background. We would be using the various allocators rather than
xmalloc and we should be adding states rather than creating another CacheVC.
The write can occur after calling back the user, so there will be no additional
latency.
4) The code doesn't seem to have provision for handling multi-fragment
documents. There are some tradeoffs there, but in any case the issues needs to
be considered. Handling of multi-fragment documents requires tighter control
over when the write is done and if and when it is successful. Again, this
would indicate we should be doing the write on the same CacheVC in additional
states.
5) The code needs to deal collisions after the initial directory is looked up.
Again, this would be easier if the SSD code was operating in the same CacheVC.
Please think about these comments. Let's use http://codereview.appspot.com/
for detailed reviews. I have already added traffic server as a repository. I
have imported the patch as well, but I think we still have some design
discussion to do before we can get into the details.
> Support ssd
> -----------
>
> Key: TS-745
> URL: https://issues.apache.org/jira/browse/TS-745
> Project: Traffic Server
> Issue Type: New Feature
> Components: Cache
> Reporter: mohan_zl
> Assignee: mohan_zl
> Attachments: ssd_cache.patch
>
>
> A patch for supporting, not work well for a long time with --enable-debug
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira