[openstack-dev] Swift global cluster replication latency...

Shyam Prasad N nspmangalore at gmail.com
Tue Aug 19 07:01:55 UTC 2014


Hi Clay,

Thanks for the reply.
I'm just experimenting with this feature in between two server machines at
different locations.
I'm not looking for synchronous replication in between the sites.
Asynchronous replication is fine.
It would be nice if the asynchronous replication were to happen at
predictable intervals, so that I can then estimate more confidently the
maximum age of objects (most of the time) on the remote server.

Haven't explored the policy feature yet.
Going through this document now...
http://docs.openstack.org/developer/swift/overview_policies.html

Will read it and report back if that is what I need.

Regards,
Shyam


On Tue, Aug 19, 2014 at 12:20 AM, Clay Gerrard <clay.gerrard at gmail.com>
wrote:

> Correct, best-effort.  There is no guarantee or time boxing on
> cross-region replication.  The best way to manage cross site replication is
> by tuning your replica count to ensure you have primary copies in each
> region - eventually.  Possibly evaluate if you need write_affinity at all
> (you can always just stream across the WAN on PUT directly into the primary
> locations).  Global replication is a great feature, but still ripe for
> optimization and turning:
>
> https://review.openstack.org/#/c/99824/
>
> With storage policies you now also have the ability to have a local policy
> and global policy giving operators and users even more control about where
> they need their objects.  For example you might upload to local policy and
> then manage geo-distribution with a COPY request to the global policy.
>
> Do you have a specific use case for geo-distributed objects that you could
> share or are you just trying to understand the implementation?
>
> -Clay
>
>
> On Mon, Aug 18, 2014 at 3:32 AM, Shyam Prasad N <nspmangalore at gmail.com>
> wrote:
>
>> Hi,
>>
>> Went through the following link:
>>
>> https://swiftstack.com/blog/2012/09/16/globally-distributed-openstack-swift-cluster/
>>
>> I'm trying to simulate the 2-region 3-replica scenario. The document says
>> that the 3rd replica will be asynchronously moved to the remote location
>> with a 2-region setup.
>>
>> What I want to understand is if whether the latency of this asynchronous
>> copy can be tweaked/monitored? I couldn't find any configuration parameters
>> to tweak this. Do we have such an option? Or is it done on a best-effort
>> basis?
>>
>> Thanks in advance...
>>
>> --
>> -Shyam
>>
>> _______________________________________________
>> 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
>
>


-- 
-Shyam
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140819/bc0b8ed8/attachment.html>


More information about the OpenStack-dev mailing list