Open Stack

Motonobu,

In our current (very initial) version of design, the proxy-server may know
about it's affinity to particular region (i.e. this is parameter set in
proxy-server.conf file) and prefers to read data from nodes in 'local'
region when serves GET request. This is implemented as splitting 'nodes'
array into 'local' and 'foreign' and shuffle them separately, then concat
these lists back with 'local' on top of 'foreign'. PUT requests are served
as usual, this is to be addressed in following iterations.
The replicator enhancements are also targeted in following iterations.
We're still finalizing design of replication and will greatly appreciate
any suggestions.

--
Best regards,
Oleg Gelbukh
Sr. IT engineer
Mirantis Inc.


On Fri, Nov 30, 2012 at 10:40 AM, Motonobu Ichimura <motonobu at gmail.com>wrote:

> Hello,
>
> I'm very interested in this topic.
>
> As you might know, we'd implemented colony(
> https://etherpad.openstack.org/grizzly-colony) for that purpose,
> and our modification to Swift is here.
>
>
> https://github.com/nii-cloud/swift/commit/173c78bed7b28ff4dbe5aa6d7063ef69e12772ef
>
> * our current implementation doesn't modify ring's data structure, so that
> we cannot store objects across specific regions.
>
> so, I just want to know that how swift-proxy/object-replicator use this
> proposed data structure.
> Is there any idea?
>
>
> --
> Best regards,
> Motonobu Ichimura
> NTTDATA INTELLILINK
>
>
> 2012/11/22 Oleg Gelbukh <ogelbukh at mirantis.com>
>
>> Hello,
>>
>> As a part of our work to implement multi-region cluster for Swift, we'd
>> like to run our design proposal by the community. We greatly appreciate any
>> comments. We're already in early stage of coding this feature, and we
>> really want to adopt community feedback at this stage.
>>
>> Design paper:
>>
>> https://docs.google.com/document/d/18VM56iSsazHbgCdbzM_aYvrhCCHKn81jmeRfsNRS7cg/edit
>>
>> --
>> Best regards,
>> Oleg Gelbukh
>> Sr. IT engineer
>> Mirantis Inc.
>>
>>
>> _______________________________________________
>> OpenStack-dev mailing list
>> OpenStack-dev at lists.openstack.org
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>>
>
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20121206/10e4a5e9/attachment-0001.html>