HPE Nimble Storage SFA. Whats that about then? Part 6 – Putting your backup data to work.


Nimble SFA – Putting your backup data to work.

Following on from part 5 – a closer look at deduplication of this blog series, we will take a look at how the Nimble SFA helps to put your backup data to work.

What on earth am I talking about?

For me at least, it seems to have become a trend in the industry that simply storing ones backups in case of a disaster is not good enough anymore. If you think about, a backup in its simplest form will likely involve a process of copying data from one location to another and storing it in a propriety format. After that, the data just sits there until such a time that you may need to recover from the backup as the primary source of the data has an issue. If you are fortunate, the backup data is likely going to sit and do nothing until it has reached its retention policy and it is deleted again.

In the case of backup software like Veeam, those backups are very likely to contain a full copy of your production virtual environment, be that VMware or Hyper-V. Imagine what you could do if those backups could be brought online as virtual machines in an isolated sand box environment.

OK, the idea of the sandbox environment is not something new, a lot of virtual machine backup vendors offer this kind of functionality these days. Veeams implementation of this is built on a technology they call Instant VM RecoveryHigh level, Instant VM Recovery allows you to present NFS storage to the hypervisor environment from the Veeam backup and replication server. Veeam then publishes a virtual machine direct from the backup file which can be brought online just like any other virtual machine.

Cool, how does the Nimble SFA fit in?

With the above in mind, the instant recovered virtual machine will be running from and served IO from the backup storage. The performance of the virtual machine is largely governed by the performance of that storage. Trying to boot a virtual machine from a traditional deduplication device is painful as this is not what its intended use case is. Rehydration of deduplicated data can take a long time. The Nimble SFA however states that on the SF100 array, it can deliver 20000 IOPS, moving up to 40000 IOPS on the SF300. Compare that to a 4 bay NAS enclosure with 7.2K SATA drives. At best that is going to be delivering about 300 IOPS in RAID 10 with 90% read bias. I use the NAS example as I have seen a lot of it out there for use as a primary backup target.

The Nimble SFA is delivering primary storage IOPS capabilities from deduplicated storage. Performance and deduplication is not something that used to go hand in hand.

Testing

To demonstrate the capabilities of the SFA, I will compare the boot time of a vanilla 2012 R2 VM from an all flash vSAN VS boot time from a backup on the SF100. Both VM instances will be booted from the same host so the only difference is the storage.

Boot time of test VM from all flash vSAN from clicking power on to login screen – Production Storage. Time: 7 seconds

boottime_vsan

Boot time of test VM from Nimble SF100 from clicking power on to login screen – Backup Storage. Time: 11 seconds

boottime_sfa

 

Further use cases with Veeam

We have proved that the SF100 is more than capable of handling an Instant VM Recovery. Veeam SureBackup builds on this technology to validate backup sets by performing an instant VM recovery and verifying that certain functions in the virtual machine work such as VM tools has launched and a ping response can be received from the VM. Customers tend to shy away from this feature as it puts a strain on the backup storage. With the Nimble SFA range, this is not a concern anymore. The SureBackup application groups can also remain online to allow for tasks such as data warehousing or testing application upgrades before rolling them out in production.

Conclusion

If performance and space-saving are something you are looking for in a backup repository, you could do a lot worse than the Nimble SFA range. Something I did not touch on is that if you have an existing investment in Nimble for primary storage, all the SAN to SAN replication features works as they would replicating between two hybrid or all flash arrays. The SFA can be used as a storage replication target and could be used with applications such as VMware Site Recovery Manager for full DR fail-over scenarios.

 

Please check out the other posts in this blog series.

Ian

Leave a comment

Your email address will not be published. Required fields are marked *