K
K
kodi2015-09-04 15:49:14
Computer networks
kodi, 2015-09-04 15:49:14

How to store data for processing in a virtual machine on a storage system?

There are nodes for virtualization, there is storage.
Images of virtual machines lie on the storage system and are loaded from it.
The question arises how to organize the storage of data with which virtual machines work.
First option:
Store data inside the image along with the OS and pull the image over iSCSI.
The second option:
the OS is in the image and competes for iSCSI. And store the data on the storage system and pull it over the network via NFS or SMB.
Which option is more productive? Or is there something more correct?
Thank you.

Answer the question

In order to leave comments, you need to log in

3 answer(s)
I
Ivan, 2015-09-04
@LiguidCool

Well, iSCSI will probably be more productive. But in general, it depends on the data, maybe NFS or SMB is enough. In addition, it is more difficult to rummage between machines.

A
athacker, 2015-09-04
@athacker

Depends on the amount and type of data. If you have a mega-archive there, or you need high performance, it's better to create a separate LUN and connect it via iSCSI - there will be less overhead when working with the volume. In other cases, it is better to use virtual disks. They are easier to maintain - backup, resize, migrate.

A
Alexey Cheremisin, 2015-09-04
@leahch

Images of virtual machines via iSCSI (we really do this directly through ceph), the rest, for general access, via NFS.
We make a common image with the system, and make snapshots from it for each virtual machine instance. We give the snapshot to the virtual machine as a block device. If in some instance you need a separate disk with data, then we give it separately. We also often make block devices for large temporary files, we also attach a separate disk to the instance.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question