NetApp 2040 Redundancy

[STORAGENOOB]

Ok, I am very new to NAS/SAN and especially NetApp.

The company I work for has a 2020, which needed a firmware upgrade (from my understanding), which caused down-time for the 2020.  (This is being retired soon)

So we are going VMWare next year and already have a new 2040 with 2 shelves full of disks, for VMWare storage.

We are purchasing an identical 2040 with 2 shelves full of disk for redundancy. 

My question/statement is: the current 2040 has 2 active controllers. So what are we gaining with the purchase of the second 2040?  I feel I am missing something, but I'm comming in to the project half-way through, and the original project person is not available to clarify the reasoning of this.

I would also like to implment a SQL Cluster, and was looking at a NetApp to fulfill that requirement as well. I can not have any databases perform poorly due to I/O.  So, am I looking at another NetApp (2040) to take on that responsibility?  Its a toss-up if it will be approved, but I need to put number together for next year's budget, and the SQL Cluster could double in price if I need to purchase an additional NetApp?

BTW - I also put in the budget for NetApp training for next year, so hopefully I won't have such basic questions next time.

Thanks

[scottgelb · Accepted solution]

Good questions... the 2 nodes are active/active and also failover for each other.  So if node2 dies, node1 takes over node2 and serves both workloads and vice-versa.   Note that your 2020 and 2040 use the same 2u chassis and the same internal disk drives, so you could controller swap between them also.  For performance, a sizing exercise is needed to look at existing workload then the addition of your SQL workload.  It may or may not work and your NetApp SE or VAR can run through the sizing tools to get a better idea.  Some data collection will be needed for this.   The FAS2040 is the entry level controller but it isn't a slouch...but not a high end 500MB/sec 60K IOP/sec controller either so the best answer is it depends. 

For classes...the ncda boot camp might be a good start and is offered by netapp and training partners.

[STORAGENOOB]

Thanks Scott.

So having 2 - FAS2040's for one vCenter farm (5 hosts) is ???

- Required?

- Recommended?

- Complete Overkill?

- Failover to a Failover?

I'm trying to understand why we are purchasing a second one (strictly failover was the reason that was given to me)?

Is there any reason that one 2040 would be completely unavailable (such as software upgrades, firmware upgrades, our datacenter being destroyed ) ?

[scottgelb]

2 controllers let's you load balance for performance. And failover/giveback to perform maintenance. If you are talking a completely separate system that could be remote DR using snapmirror and srm or snapvault for a backup.

Typos Sent on Blackberry Wireless

[paulstringfellow]

Scott has pretty much nailed all that for you…

The 2040HA is a single device with two controllers installed into it…so a 2U chassis with both controllers installed.

This provides local resilience as well as the ability to do things like non disruptive updates to controllers, or balance load etc.

In terms of a second 2040, then as suggested this would be for DR purposes, so this is down to a major failure of production array, so as you said, loss of production data centre… so you can snapmirror the data to another location….and using manual process or tools like vmware SRM then you can recover your vm’s elsewhere.

In terms of SQL performance, then again he is spot on, talk to your partner or local NetApp sales rep about getting some performance stats from your current SQL environment. this will tell you as to whether you need a separate 2040, more spindles (so another SAS shelf maybe into your 2040 production box) or even that the current spec you have will give you all the performance you need.

Hope that helps…as I say Scott pretty much nailed all that for you…