Ask Your Question
0

Does the Block and Object storage are obligatory?

asked 2014-12-13 11:32:59 -0600

AlonFre gravatar image

Hi all, we are following the "OPENSTACK INSTALLATION GUIDE FOR UBUNTU 14.04 - JUNO" guide, we have a question regarding the Block and Object storage nodes.

In this installation guide, the storage nodes are optional, but if we dont have extra nodes for the Block and Object storage, we must install them in other nodes? (for example the controller, compute1 or network nodes), or they are not obligatory components?

Thanks, Alon

edit retag flag offensive close merge delete
add a comment

1 answer

Sort by ยป oldest newest most voted
1

answered 2014-12-13 12:20:38 -0600

Mzoorikh gravatar image

About block level storage(Cinder), two different services need to be installed on your machines:

  • Block Storage API and Scheduler: Installed on Controller node
  • Volume Service: Installed on storage node(Storage nodes are one or more server with directly attached storage or SAN connectivity that provides volume access for openstack instances)

If you don't have extra nodes for volume service, you have to install volume service on one of your openstack nodes that has enough local attached storage disk to share(e.g. compute, controller, network).

The case is similar for object storage(Swift).

Both block(Cinder) and object(Swift) storage services are completely optional. You can install openstack without object storage. But if you want to install openstack without Cinder, you have to install nova-volume service. Becasue your instances need volume service. Nova-volume is openstack legacy volume service that provides simple volume interface for openstack compute nodes.

edit flag offensive delete link more

Comments

1

You do not need Cinder or nova-volume (nova-volume isn't even a service anymore, it IS Cinder). Instances do _not_ need a volume service.

SamYaple gravatar imageSamYaple ( 2014-12-13 22:44:28 -0600 )edit

Hi SamYaple, Can you please elaborate more? If instances do not need a vakume service, where their own data is kept? Is it just on the compute node where they reside? Thanks (:

AlonFre gravatar imageAlonFre ( 2014-12-14 00:25:25 -0600 )edit
1

Yes AlonFre. The data is stored on the compute node where the instance resides by default. It is ephmeral data, it goes away when the instance stops. Cinder is used for persistent data.

SamYaple gravatar imageSamYaple ( 2014-12-14 02:24:41 -0600 )edit
add a comment

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

subscribe to rss feed

Stats

Asked: 2014-12-13 11:32:59 -0600

Seen: 191 times

Last updated: Dec 13 '14

Related questions

complete diagram showing the object storage architecture, which includes the proxies, rings, zones, partitions, storage nodes, objects, account database, container database ...

What configurations are needed to enable boot from volume?

Are there distance limitations in physical OpenStack architecture ? [closed]

defining storage node for openstack swift

international standard for object storage architecture [closed]

Can i upload large dataset to block storage before attaching it to the instance?

S3 commands to swift via swift3 middleware unauthorized 401 Issues

Direct link to Swift objects

How can I get the IQNs of iSCSI initiators in OpenStack?

How much memory on block storage is required?