[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Openstack] EC stability status in openstack swift


Erasure codes are widely deployed and used with Swift. You're right that EC is definitely more expensive, though. It will take more CPU and quite a bit more network management (more connections). Interestingly, the CPU overhead for the EC algorithms isn't all that much (small single-digit percentage differences). The biggest costs come from the increased number of network connections and having the OS manage that. For example, with 3x replicas, you've got 3 sockets to manage for writes and one for reads. For a 10+4 EC scheme, you've got up to 14 for both reads and writes. (Also make sure you're only using EC for objects that are bigger than 1MB.)

--John




On 23 Mar 2018, at 0:13, Shyam Prasad N wrote:

> Hi,
>
> I wanted to understand the stability of the erasure coding feature in
> openstack swift. Is it in widespread production use?
> We're exploring the possibility of using EC instead of replication on
> few of our servers which are not so loaded in terms of storage
> capacity.
> Also, I expect EC to be more CPU intensive than replication. Is there
> some recommended minimum configuration to support EC?
>
> Thanks in advance.
>
> -- 
> -Shyam
>
> _______________________________________________
> 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 --------------
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/20180323/9d16f35b/attachment.sig>