Open Stack

For example, you can have 3 replicas stored in a global cluster and get dispersion across multiple geographic regions. But it's all one logical cluster.

With container sync, you've got separate clusters with their own durability characteristics. So you would have eg 3 replicas in each cluster, meaning 6x in the data that is synced between two clusters.

--John



On 18 Feb 2018, at 22:11, aRaviNd wrote:

> Thanks John.
>
> You mentioned sync process in global clusters is more efficient. Could you
> please let me know how sync process is more efficient in global clusters
> than container sync?
>
> Aravind
>
> On Wed, Feb 14, 2018 at 9:10 PM, John Dickinson <me at not.mn> wrote:
>
>> A global cluster is one logical cluster that durably stores data across
>> all the available failure domains (the highest level of failure domain is
>> "region"). For example, if you have 2 regions (ie DCs)and you're using 4
>> replicas, you'll end up with 2 replicas in each.
>>
>> Container sync is for taking a subset of data stored in one Swift cluster
>> and synchronizing it with a different Swift cluster. Each Swift cluster is
>> autonomous and handles it's own durability. So, eg if each Swift cluster
>> uses 3 replicas, you'll end up with 6x total storage for the data that is
>> synced.
>>
>> In most cases, people use global clusters and are happy with it. It's
>> definitely been more used than container sync, and the sync process in
>> global clusters is more efficient.
>>
>> However, deploying a multi-region Swift cluster comes with an extra set of
>> challenges above and beyond a single-site deployment. You've got to
>> consider more things with your inter-region networking, your network
>> routing, the access patterns in each region, your requirements around
>> locality, and the data placement of your data.
>>
>> All of these challenges are solvable, of course. Start with
>> https://swift.openstack.org and also feel free to ask here on the mailing
>> list or on freenode IRC in #openstack-swift.
>>
>> Good luck!
>>
>> John
>>
>>
>> On 14 Feb 2018, at 6:55, aRaviNd wrote:
>>
>> Hi All,
>>
>> Whats the difference between container sync and global cluster? Which
>> should we use for large data set of 100 Tb ?
>>
>> Aravind
>>
>> On Feb 13, 2018 7:52 PM, "aRaviNd" <ambadiaravind at gmail.com> wrote:
>>
>> Hi All,
>>
>> We are working on implementing Openstack swift replication and would like
>> to know whats the better approach, container sync or global cluster, on
>> what scenario we should choose one above the another.
>>
>> Swift cluster will be used as a backend for web application deployed on
>> multiple regions which is configured as active passive using DNS.
>>
>> Data usage can grow upto 100TB starting with 1TB. What will be better
>> option to sync data between regions?
>>
>> Thank You
>>
>> Aravind M D
>>
>>
>> _______________________________________________
>> 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/20180220/74bcc478/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: OpenPGP digital signature
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20180220/74bcc478/attachment.sig>