we have been long time HP shop but now we are considering using NetApp products for unified storage. Reasons are CIFS, iSCSI, FC, NFS on one box. Simple as that. But there is a lot of FUD between EMC, HP towards NetApp and vice versa and I am really getting a bit confused. So here I would like to ask for your opinion on this NetApp stuff:
how to actually calculate usable storage if FAS is configured for max protection/max perfomance ? Let's leave shapshoots aside. I may need few of them but not that many. Please do not start with depends
is perfomance good enough if we use FAS2020A for CIFS file server of user's home dirs and departments data ?
this will be used by c-3000 enclosure from HP so I would ilke to ask is there anybody with similar config ?
Thanks for considering a NetApp array. It is very inderstandable that your current storage vendors woudl like to and are working very hard to retain your business. With that said it is not in their best interest to present a strong case for NetApp. Maybe I can help. I've been with NetApp for 10 years andwas a cusotmer for 2 years prior to joining...
On 'how to calculate usable storage' - all hard drives are 'right sized' - this occurs with NetApp, HP, EMC, etc... Blame drive vendor for not getting 1 TB usable out of a 1 TB drive. NetApp uses RAID-DP whihch allows for the simultaneous failing of any two disk drives in a RAID group. The default RAID over head is 2 disks for every 16 (12.5% RAID overhead). One or mor RAID groups create an aggregate. Aggregates reuiqre a 3% reserve if you enable data deduplication. From there, one has their total capacity.
On 'can a FAS2020A be used for CIFS file server' - This depends on how many users you have. I'd suggest that a FAS 2020A could serve a couple thousand users.
On 'support for a HP C30000' - the C3000 is a blade chasis. The supportability would be determined by the type of connectivity. Can you elaborate?
Thansk for considering us, there's a reason why NetApp cusotmers are referred to as 'NetApp fan bois' by other storage vendors... Our tech is pretty cool and very different than HP & EMC.
being independent gives you choice to review all offering. NetApp wasn't option for us few years ago but now solely it is. There are many issues related to support in our country (complicated import/export procedure). Let me stress adition questions:
@'how to calculate usable storage': let's said I have 12 x 300 GB in FAS2020A. What exactly storage usable I will get from it ? Stress really numbers. Usable means I can save something there
@ 'can a FAS2020A be used for CIFS file server': Ok, very nice feature to have redundant file server
@ 'support for a HP C3000': We can use just about anything. From L2/L3 drives for iSCSI to direct FC or NFS. It depends on configuration and money customer is willing to pay
So, you are still up to give me real numbers. Since your 12.5% doesn't come with some other post I read on NetApp blogs.
I came from an EMC background and we just purchased a NetApp FAS2020a here. I must say, NetApp knows what they are doing.
NetApp has a particular kind of raid setup on their SANs called RAID-DP. so you lose a few more disks for that added protection, but also regaining alot more disk space because of de-dup on the primary storage.
we have 12x300 SAS drives in our FAS2020a and this is the configuration we used to get the most diskspace possible. I would suggest getting the 450gb SAS drives. I've even heard that you can use SATA drives and not see a performance hit, but I wasn't bound to let my vSphere implementation possibly suffer just to save a few thousand bucks.
Controller 1 - 9 disks are used for this "Aggregate". 2 are used for parity, 1 used as a spare. so you lose 3 drives of space, giving total of "usable" space as 1.36TB. but even that is removed some because 10-20% of that will be used if you plan on using SnapShots on teh SAN.
Controller 2 - 3 disks are used for this "Aggregate". 2 are used for parity, no spares are used (thus we get an error saying it's an incorrect configuration, but it still works and we don't have to worry about losing Data because there are 2 parity drives. NetApp has a phone home feature, so if a drive does fail, there is one here the next day to start rebuilding the RAID), giving us a total of "usable" space of 275GB.
It doesn't sound like a lot, but believe me, after de-duplication, we still have room to spare with a total of 15 servers so far. I would suggest getting the 450gb drives or perhaps getting a demo of SATA to save more money.
Also, whatever Vaughn preaches about VMware and NetApp is basically gospel. Vaughn and some others wrote the NetApp and VMware vSphere best practices guide. get yourself familar with it.
Thank you for reply. So out of 3,6TB you are getting something like 1.5TB of usable space? If this is true, then some "fud" is not fud. However, I am fully aware that you have to give something up to get something back. If we are giving space for nice features, then it is Ok but still NetApp should work this out without dedup :-). I have not doubts about features of NetApp only perfomanse (which is in small enviroment more then enough) and usable disk space. If I expand FA2020a, I have to do it with FC disks (SATA is not an option) and this is raising cost high.
Consider the 12x300GB FAS2020A array
The controller assigns drives to each controller on a drive by drive basis.
controller-1 w 9 drives (2 RAID) (1 spare) (6 for data)
controller-2 w 3 drives (2 RAID) (1 for data)
It appears that controller 2 is configured as a ‘hot’ stand by, which is an OK config. I would suggest reducing the RAID-DP to RAID4 on controller 2, and configuring the freed up drive to controller 1.
controller-1 w 10 drives (2 RAID) (1 spare) (7 for data)
controller-2 w 2 drives (1 RAID) (1 for data)
The utilization numbers look poor here, as the total number of drives is rather low. Consider doubling the number of drives from 12 to 24…
controller-1 w 12 drives (2 RAID) (1 spare) (9 for data)
controller-2 w 12 drives (2 RAID) (1 spare) (9 for data)
In this last config you’ll notice the RAID overhead remain the same is in the original, yet the amount of addressable storage is increased.
24 total drives
18 data (75%)
4 RAID (16.7%)
2 spares (8.3%)
Now if you reserve space for snapshots cap will decrease, if you enable dedupe cap will increase.
You could also reduce RAID-DP to RAID4 (if data protection is less of a concern). This would buy you two more data drives.
I don't work for NetApp, but I have a feeling that they will strongly encourage you to stay with RAID-DP. That is the RAID configuration they use to make sure you don't lose data because of failed drives.
I agree RAID-DP is the preferred way, but when thin on drives it makes sense. Consider a two drive RAID4 aggregate actually behaves like a RAID1 mirror. Having more parity drives than data isn't optimal.
As for shelf expansion you will need to see your sales rep as I really focus on integration and as such I slip on hardware capacities and the like. I work in the virtual space... what's hardware?