Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1108439

Summary:	[RFE][swift]: Efficient Replication for Global Distributed Cluster
Product:	Red Hat OpenStack	Reporter:	RHOS Integration <rhos-integ>
Component:	openstack-swift	Assignee:	Pete Zaitcev <zaitcev>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Mike Abrams <mabrams>
Severity:	unspecified	Docs Contact:
Priority:	high
Version:	unspecified	CC:	derekh, mabrams, markmc, nlevinki, panbalag, scohen, sgotliv, srevivo, thiago, zaitcev
Target Milestone:	z5	Keywords:	FutureFeature, Reopened, Triaged, ZStream
Target Release:	7.0 (Kilo)
Hardware:	Unspecified
OS:	Unspecified
URL:	https://blueprints.launchpad.net/swift/+spec/efficient-replication
Whiteboard:	upstream_milestone_none upstream_definition_new upstream_status_unknown
Fixed In Version:		Doc Type:	Release Note
Doc Text:	Efficient replication of distributed regions is now possible. Existing replication sends data over unnecessarily. For example, in case of two regions and replication factor of four, a typical spread is to have two replicas per region, assuming uniform weights. If the long-haul link goes down and up, replicators send two copies of every outstanding object across regions. This is not necessary. If only one copy is sent over, replicators of the remote region restore complete redundancy without using the long-haul bandwidth. This update ensures stable performance during a recovery after an outage in a multi-region cluster. It has no effect inside a one-region datacenter.	Story Points:	---
Clone Of:		Environment:
Last Closed:	2017-08-31 15:44:29 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description RHOS Integration 2014-06-12 04:09:33 UTC

Cloned from launchpad blueprint https://blueprints.launchpad.net/swift/+spec/efficient-replication.

Description:

I propose a more efficient way for object replication. The main idea is reducing the amount of data transfer between regions.  This blue print is involved  with a design session of juno summit. See following summit description and etherpad  for more information on this.

Session Description:
http://junodesignsummit.sched.org/event/7ae1af936b54b937a92db9c4344dfe66#.U3OKs_l_sU0
Slides:
http://www.slideshare.net/tsuyuzaki/open-stacks-ummitjuno
etherpad:
https://etherpad.openstack.org/p/juno_swift_object_replication

Specification URL (additional information):

None

Comment 2 Pete Zaitcev 2015-03-26 13:25:11 UTC

Main review:
 I5d990444d7977f4127bb37f9256212c893438df1
 https://review.openstack.org/99824
Update:
 Ie8f75ed65c7bfefbb18ddccd9fe0e41b72dca0a4
 https://review.openstack.org/157290
Update:
 Ic59d998a3e36a3eb3e509d9fdf7096e812281357
 https://review.openstack.org/159683

Comment 3 Pete Zaitcev 2015-03-29 18:38:46 UTC

(wrote a more concise doc, see "Doc Text" field in the bug)