[Openstack] Swift And Erasure Code Storage

Luse, Paul E paul.e.luse at intel.com
Wed Aug 13 21:57:30 UTC 2014

The answer totally depends on how you choose to configure EC.  We generally refer to the “extra storage” as the overhead of the durability policy.  For triple replication, obviously its 3x (you buy 3x the usable capacity you need).  For EC it depends not only on which scheme you choose but the partakers that you configure.  For example Swift will support a few different Read Solomon schemes out of the box (when its done) and from there you can choose the ratio of data:parity such as 10:4 where you’d have 14 total disks, 10 of them for data and 4 for parity.  In this scheme you could lose 4 disks and still recover your data and your overhead would be 1.4 (14/10) as opposed to triple replication of 3 (3/1)


From: Brent Troge [mailto:brenttroge2016 at gmail.com]
Sent: Wednesday, August 13, 2014 2:41 PM
To: openstack at lists.openstack.org
Subject: [Openstack] Swift And Erasure Code Storage

With a 100% 'Erasure Code' policy how much extra storage is needed to satisfy a 1PB usable cluster ?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20140813/5147a413/attachment.html>

More information about the Openstack mailing list