Hi guys,
I work for a small business (~100 employees) and we're stepping into the heady world of SANs. We've been doing some VM consolidation for a while, and hosting this from standard servers with locally attached disks and software iSCSI terminators/NFS. We've been burnt a bit on investments on basic RAID arrays that didn't work out well (the vendor shall remain nameless!) so I think I can make the case for getting a platform that's a bit more stable.
We are considering purchasing a FAS2020. I'm attracted to this product in particular because it supports iSCSI and CIFS/NFS in the same box. The primary function will be to serve XenServer and Hyper-V virtual machines, and I'm also hoping that we'll be able to boot the Xen hypervisor (and our other non-virtualized servers) over iSCSI, so our servers will be diskless. I'm also hoping that we can serve standard NFS and CIFS volumes from the same FAS2020. In other words, I want to try to consolidate all my storage in our small "datacentre" into one place. At the moment there are about ten VMs and around about 20-25 servers which we may be able to consolidate. We've currently got about 3TB of data being served out over NFS/CIFS, but the server workloads are very light indeed - really just documents, images, etc.
We don't really need XenServer Essentials because our VMs tend to be long-lived (we're not regularly creating/copying/deleting the VMs), so I'm happy enough to create a couple of standard iSCSI or NFS SRs, and create volumes manually within that in the way we do right now. We have an add-on script that snapshots running VMs and copies the backup off to another server where it is streamed to tape.
It's important to me that the investment will stand for as long as possible. With the "other vendor" we were sold what turned out to be an end of line device that used PATA disks. High density PATA disks are getting hard to find. The array obviously has no cool stuff such as dedup/snapshots/etc.
I've been busy researching the FAS2020 capabilities and have a few questions. I'd be delighted if I could obtain clarity on these matters as these will help me put a sound business case to my management.
(before we get into that, are there any reputable resellers of this equipment in the UK ? I've emailed several suppliers who appear on NetApp's UK reseller list, with details of my requirements, storage needs etc, but not one of them has replied or attempted to call me! That's kind of surprising in a recession .. if you are a reputable seller and you're reading this, please reply with your details and I'll be in touch)
1. What's the difference between the FAS2020 and the FAS2020A ?
2. Is it true that each controller requires three disks for it's own use and, therefore, in an active-active configuration with dual controllers, six out of the total 12 disks will not be available for storage ? If so, what is the reason for this ?
3. If (2) is true, is it therefore possible to use small/cheaper disks for the three required for each controller, and use larger/higher performance disks for the array itself ?
4. Is it true that each controller must have at least one Aggregate assigned to it, and therefore in an active-active configuration I must create two Aggregates ?
(the alleged limitations with dual controllers are leading me to believe that it would be better for me to get a pair of FAS2020s and have them mirrored, failing over in the event of a controller failure - which is something we would probably want to do in the event of controller failures anyway)
5. Are hot spares global across the whole appliance including any expansion shelves, or are they per-Aggregate ?
6. I assume that I cannot use commodity disks in a NetApp caddy, and that I must use NetApp-branded disks ?
7. It looks like the FAS2020 is approaching EOL and that it is being heavily discounted. Can I expect that I will be able to obtain new disks and shelves for this product well into the future ? I'd expect to get at least five years of service life from it.
8. If I create a Snapshot, is the snapshot allocated out of the storage assigned to the FlexVol ?
9. If I understand correctly, my configuration steps would be :
(a)Create an Aggregate, which is NetApp's term for a RAID volume. If I understand correctly, it looks like RAID-DP is the way to go. A disk can obviously be present in only one Aggregate at any one time.
(b) On top of the Aggregate I can create as many FlexVols as I like, up to a reasonable limit, which can be of any size, and which can be overcommitted. I must keep the FlexVol to a maximum of 1TB to take advantage of de-duplication.
(c)The FlexVol can then be served out as either an iSCSI target or an NFS volume. I can then configure XenServer to use this as an SR.
Have I got this right ?
10. What's the deal with OS upgrades; is this free or is there an extra charge for newer functionality ?