We have recently become a Nimble Partner, and after years of supporting Netapp arrays, it was nice to see a very fresh approach, ‘storage can be simple’. I’ve been managing Netapp environments for over 5 years, and like the concepts Netapp bring to the table. Get a bucket load of disks, group them together to create one performance pool and carve off slices for all you applications. As a concept this is great, we all love the fact that IT now has this great big pool of storage and we can carve it off when ever we want.
So when Nimble came along with the similar concepts and easier management, we thought, this looks cool, and its going to have a huge affect at the way everyone manages their storage. Gone are the days were you need to understand aggregates, Qtrees, relocate commands, CIF time outs, CF Failover options, and Active/Active capacity management. Nimble makes it simple.
Netapp’s Qtrees are one of the things I never really got, to me its just another layer that needs to be created inside your Netapp, in order for certain functionality to work. I’ve spent hours trying to figure out why OnComand didn’t like my snap mirror volume; only to figure out I didn’t have any Qtrees in place. Setting up replication with Nimble has made my life so easy; with Nimble all it takes is four simple steps:
- Create your partner array – 1 minute
- Select your volume – 1 minute
- Select your replication schedule – 1 minute
- Monitor and grab a cup of coffee.
Whenever we deploy a nimble array we always upgrade the controllers, to get the longest support and the most functionality. This process is extremely straightforward: click to download the firmware; upload it to the array and then click to upgrade. The system will automatically patch the passive controller – bring it up then failover and patch the new passive controller – all without an outage. In fact, Nimble support have seen a trend where customers are upgrading their arrays during the middle of the day.
I love this process because I don’t have to read a 30-page document to find out that I have to flash all my disks, upgrade all the firmware on my network cards and all this just so I can get my CIFS snaps to work again in windows 7.
Active/Active v Active/Passive
One of the questions we all ways get is why not Active/Active? It does have its benefits – you can carve up your disks and split one side for your CIFS, backup workload and have the other for your core VMWare and SQL workloads. This works really well, giving your core infrastructure the resources it needs to do the grunt work while your backup server uses another set of disks. But when there is a -power -network -fibre channel -grumpy kernel issue that causes a controller failover it can be a very painful 30-120 seconds of ‘where is my storage’, leaving applications asking where did my storage go. We can of course change time out values to get around these issues, but ask any help desk about a storage controller failover and you will not hear a ‘happy’ story.
Active/passive- Yes you do have an inactive controller sitting there, but because of this failovers are quick, as you don’t have to wait for CPU time that was processing your backup request when you controller failed. In the Nimble world it works like this: controller fails – start passive mirrored controller. Simple.
What Netapp has that Nimble doesn’t
CIFS – I’ve learnt to live with out this – it’s just as easy to go back to the good old days of a windows server with quotas
NFS – The reason we love it is that it allowed VMware to talk directly to the SAN to allow for grouped snap shots of your VM’s. Nimble’s VMware plugin provides the same functionality allowing you to sync your snap shots at the same time.
API – I’ve got a lot of scripts for creating and managing Netapp arrays – but what I’ve figured out is thanks to the simplicity I don’t need these any more. I still get a CLI so worst case is a shell script that creates my volumes and sets my replication up if I still want/need to script it.
Dedupe – this was the hardest part – till I realized the benefits of compression
As I mentioned earlier, I’ve recently spent a fair amount of time setting up Nimble array’s and because it is so simple, I have had time to get on with the fun stuff: like fixing Exchange DAG fail over issues or working out why your Cisco Call manager server can’t phone Albury from your Perth office. The things that a Netapp can do are great, but Nimble does it better because they make it simple. Leaving us to get on with our lives and enjoy our work free weekends, now that part is awesome.