Bug 1041035

Summary:	[RFE][nova]: Loadbalancing for instance storage paths
Product:	Red Hat OpenStack	Reporter:	RHOS Integration <rhos-integ>
Component:	RFEs	Assignee:	RHOS Maint <rhos-maint>
Status:	CLOSED UPSTREAM	QA Contact:
Severity:	low	Docs Contact:
Priority:	unspecified
Version:	unspecified	CC:	markmc, yeylon
Target Milestone:	---	Keywords:	FutureFeature
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
URL:	https://blueprints.launchpad.net/nova/+spec/nova-instance-storage-loadbalancing
Whiteboard:	upstream_milestone_none upstream_status_not-started upstream_definition_obsolete
Fixed In Version:		Doc Type:	Enhancement
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2015-03-19 17:21:23 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description RHOS Integration 2013-12-12 13:28:03 UTC

Cloned from launchpad blueprint https://blueprints.launchpad.net/nova/+spec/nova-instance-storage-loadbalancing.

Description:

The default path of instance location is in /var/lib/nova/instances/#InstanceID'/

If you want LiveMigrations you are advised to have this directory on a centralized storage, for example on NFS - as explained here: http://docs.openstack.org/trunk/openstack-compute/admin/content/configuring-live-migrations.html

There might be good reasons for having more than 1 instance directory for VM deployments - for example:
- In a big lanscape you might want to have more than 1 central storage mounts, i.e. 2 NFS mounts
- In compute-nodes with multiple local raid arrays you want to distribute instances over all arrays

If you want to use more than 1 location you need:
- more directories below /var/lib/nova/instances/ with UniqueIDs
- a patch for Openstack to deploy on the directory with the least utilization

Even better would be to have a modular scheduler integrated so that it would be flexible on which algorithm the scheduler chooses the instance directories - f.e. one target for high IO VMs etc ... 

--> I already have a working implementation for scheduling instances over 2 directories based on utilization which I could commit.



Specification URL (additional information):

None