Hardware RAID setup advice (general storage question)

[sjg4502]

We are using DSS to present NFS volumes to ESX. The underlying RAID setup currently is:

logical drive 1 is RAID 1+0: 8 x 300 GB drives for VMDK/virtual machine files

logical drive 2 is RAID 5: 4 x 300 GB drives for backup,snapshots,sync,etc.

I was wondering if that layout makes sense in trying to separate backup,sync,etc activity from production virtual machine activity? Or, would performance be better overall if I just RAID 1+0 all disks, more spindles, and carved up the storage as needed?

So, there are 12 x 300 GB drives in the storage server. What config makes sense? I am leaning toward performance over storage space.

Thanks,
SG

[To-M]

I personally have not tested VMware and a RAID 1+0 on DSS but if the Snapshots, backup is not that much during operations I would go with the whole 12 drives on the RAID 1+0.

[sjg4502]

Ok. Since I'm new to DSS setup, how would I best use a full RAID 1+0 setup to:

1. create NFS shares to present to ESX for virtual machines
2. create snapshots, backups and then sync to a standy DSS for DR?

I guess item 2 is the biggest unknown right now. In other words, what is the best way to backup and sync the data to a 2nd DSS, while not impacting normal business production if I go the full 12x300 RAID 1+0 setup? Your response is much appreciated.

thanks,
SG

[To-M]

1. create NFS shares to present to ESX for virtual machines

Use the DSS and create a NAS Logical Volume then create a Share and enable the NFS file protocol. Store your ISO images for VMware.

2. create snapshots, backups and then sync to a standy DSS for DR?

Using another DSS use the Auto Failover feature ( see link below for more details).
Currently I have not had time to test all the other features with Snapshots and VMware.
For the Backup you can use our NAS Backup feature via a mapped Share from the NAS Logical Volume.

Library
Product Presentations

http://www.open-e.com/library/product-presentations/

2008 - Open-E Volume Replication with Failover over a LAN, December

[mhubert]

We have the same situation in our infrastructure.

We are using the first DSS for serving NFS Volumes to ESX, make a volume replication to the second DSS and create a snapshot of the backup volume every night.
So, I thought, the performance impact of creating snapshots is not affecting the production DSS.
But, I have not yet tested the consistency of the ESX images in the snapshot.

@To-M:
The auto-failover feature is only for ISCSI-Volumes, isn't it?

regards
Matthias

[To-M]

Yes the Auto-Failover feature is only for ISCSI-Volumes. We are working on Auto Failover for NAS as well - we hope to have before this year ends. NAS Auto Failover is extremely difficult to do as there are so more issues to contend with then iSCSI.

But it is coming

[MarcSchaefer]

Hi,
we're also planing such a setup and i'm wondering why we should use NAS instead of iscsi targets.

I have a test setup with 2 dss storages and 2 ESXi Servers. Iscsi is working fine until now. Have I overlooked some important points?

[To-M]

You are fine with iSCSI Target use the NAS is optional for storing files with using the NFS protocol from the DSS.

[mhubert]

We are using the first DSS for serving NFS Volumes to ESX, make a volume replication to the second DSS and create a snapshot of the backup volume every night.
So, I thought, the performance impact of creating snapshots is not affecting the production DSS.

Hi To-M!

What's your opinion of the performance issue I stated above?
Is this a correct way to handle a backup scenario with two DSS?

Thanks
regards
Matthias

[To-M]

If only creating a few Snapshots running at the same time then performance will not be an issue. Link below from our Knowledge Base on Snapshots.

http://kb.open-e.com/entry/2/

The total number of snapshots is dependent on the LVM, but 255 is a safe
number. There are, however, definite limitations as to the number of ACTIVE
snapshots:

32bit system:
- 10 per LV
- 20 per system

64bit system:
- 1 per LV
- 10 per system