[Openstack] [Swift] Question regarding global scaling

Adam Lawson alawson at aqorn.com
Fri Feb 21 21:59:20 UTC 2014


Sure Richard.

Currently we have three physical regions hosted internally - West Coast,
East Coast and Eurozone (I'm suspecting an APAC region is coming shortly).
Ultimate goal is public access. I'm currently in the process of
building/placing two proxies in each region for HA and defining one zone
per region for now. I know Swift assumes one region by default but I
haven't found the region settings yet so until I find those and manage to
segment by region, I'm relying on the default Region1 and
separately-defined zones for the time being.

Does that help? Currently everything is in a geo-disperse virtual
development environment until the physical hardware arrives.

Is that what you were wondering or actual configuration file settings? I'll
share what I can safely when I get regions sorted out but I'm not sure how
much I can anyway given the iron-clad NDA I'm operating under. Pretty sure
that even deployment strategy is considered protected IP.


*Adam Lawson*
AQORN, Inc.
427 North Tatnall Street
Ste. 58461
Wilmington, Delaware 19801-2230
Toll-free: (888) 406-7620



On Fri, Feb 14, 2014 at 3:59 PM, Richard Raseley <richard at raseley.com>wrote:

> On Tue, Feb 11, 2014 at 3:38 PM, Adam Lawson <alawson at aqorn.com> wrote:
>
>> Hola peoples.
>>
>> I'm working on a general purpose Swift deployment that needs to scale
>> globally. For example, nodes in West Coast, East Coast, EU and APAC. We
>> have a Swift PoC cluster that spans West Coast and EU and it works fine,
>> replicating using zones for now.
>>
>> For those who are scaling to that degree, are you building multiple
>> unique clusters and replicating between them somehow or using regions and
>> replicating within essentially one giant cluster and using affinity rules
>> like read_affinity and write_affinity*?
>>
>
> Utilizing the region capability of Swift is a good option depending on
> your use case (I've used a region-based configuration in one production
> deployment). Can you talk more about your data structure / access patterns
> in order to help myself (and others) give better advice?
>
>
>>
>> *Adam Lawson*
>> AQORN, Inc.
>> 427 North Tatnall Street
>> Ste. 58461
>> Wilmington, Delaware 19801-2230
>> Toll-free: (888) 406-7620
>>
>>
>> _______________________________________________
>> Mailing list:
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>> Post to     : openstack at lists.openstack.org
>> Unsubscribe :
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>>
>>
>
> _______________________________________________
> Mailing list:
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
> Post to     : openstack at lists.openstack.org
> Unsubscribe :
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20140221/620c4e4a/attachment.html>


More information about the Openstack mailing list