4

I am looking at building an new 2012 R2 with Hyper V setup, everything has to be HP hardware.

2x HyperV 2012 R2 servers running 20-30 VMs. These will be dual 6 core xeons with 128gb of ram.

2x 2012 R2 back end storage servers running SMB 3.0 shares, connected to the front end VM servers with multiple 10gb ethernet connections. Not sure on the hardware requirements for these servers so advice helpful.

2x SAS 6gb external enclosure for each storage server. Probably HP d2600 enclosures.

A couple of questions:

  • What is the best way to achieve high availability and redundancy in this setup? Should I use replicas or a HV cluster?
  • What is the best way to ensure redundancy of the storage back ends? File share clustering?
  • Would the performance of the SAS enclosure be fast enough? What kind of drives would be suitable for this? Would 7.2k SAS drives be enough or would we need faster? What would be the best raid setup? Initially there would be 12 drives expanding to 24. Would a mix of fast SAS or SSD and SATA be more efficient?

Thanks in advance, any help appreciated.

2

1) You need odd number of enclosures for enclosure awareness as with only two you're going to lose both data disk and witness disk for pool. Means cluster would be dead with 50% probability (depending on what actual enclosure would go South). So one is OK (no protection), three is OK, five is OK and four, six etc are NOT OK.

2) You need SAS JBODs interconnected to all of your servers as Clustered Storage Spaces cannot proceed w/o having both SAS ports of the particular SAS disk being connected to different servers. With dedicated SAS JBOD per server you're not going to achieve this so can create Storage Spaces but not Clustered Storage Spaces.

3) You may use inexpensive high capacity SATA spindles with cheap SATA flash drives mounted internally and something like Virtual (Native?) SAN from StarWind to mirror LUs and virtualize DAS into virtaul SAN. See:

http://www.starwindsoftware.com/native-san-for-hyper-v-free-edition

4) Re-think what you do. Run VMs from DAS and use some VM replication software (built-in or say VEEAM) to achieve quite moderate RTO and RPO with a minimal investments. Many apps have built-in clustering features w/o need in shared storage as well (SQL Server AlwaysOn, Exchange DAG etc). Can be also a faster and cheaper option. You'll still need third physical host to feed witness (SMB share is fine) but there's no need in having shared storage.

0

That was a lot of questions. I'm not able to answer them all at once, but I can share some thoughts with you, in regards to the general hardware requirements/specs.

Depending on your budget, I would consider adding some SSD cache. It's supported by the newer HP controllers (P22x/P42x/P82x). If you add enough cache, you might be able to get away with using the 7.2K drives, but I would consider using 10K SAS drives (2.5") in either a D2700 (or two) or buy the DL380p Gen8 servers with 25 internal drive bays. Then you can attach one or more D2600 enclosures for "cold" data, like backup and file shares. For performance, I would always recommend RAID10. Backup data can be on RAID6.

I don't have a lot of experience with SMB 3.0 and Server 2012 R2, so I would make a POC to uncover, which of the scenarios would suit you/your company best.

Hope this helps a little.

0

You can have high availability in that kind of configuration if you make the storage cluster highly available and configure it so that the SMB shares are "continuously available." In order to set that up, you need at least two machines in the file server cluster and you need SAS connections from each of the file servers to each of the enclosures. You need the enclosures have to be on the list of ones that are compatible with clustered Storage Spaces, and I have no idea whether that particular one from HP is. As your HP rep.

You also need HBAs in the storage cluster that either do no caching or RAID in hardware or can have that turned off, so that the physical disks in the enclosures are directly visible to the Windows file server cluster nodes.

Lastly, you need all the disks in the enclosures to be SAS disks that support SCSI-3 persistent reservations.

Once you have a highly available file server cluster, the Hyper-V hosts can be configured in a manner that is clustered, or not, to your tastes. If you do cluster the compute hosts, then the VMs will themselves be highly available.

  • Jake while it sounds like a good idea on the paper config you describe would require at least a pair of servers ($2,000 each) + basic Windows Server 2012 R2 license ($500 each). That's comparable to EMC VNX entry level (ask local VAR for a huge discount) and EMC has totally different level of support and SLAs compared to MSFT. store.emc.com/Product-Family/VNX-AND-VNXe-PRODUCTS/VNX5100/p/… – BaronSamedi1958 Apr 5 '14 at 23:10
  • ...also HP has NO certified SAS JBODs so config you recommend would be rejected by MSFT support with basically any issue reported. windowsservercatalog.com/… – BaronSamedi1958 Apr 5 '14 at 23:11
-1

What is the best way to achieve high availability and redundancy in this setup ? Should I use replica's or a HV cluster ?

Both are not capable of doing any HA.

  • Replica (assuming you talk of DFS replication, which is the only replication that is supported with the technology you name) will not replicate open files so you gain no high availability.

  • HV Clusters (assuming you mean Windows HA CLustering) relies on replicated high available storage, so you can not achieve this the technolgy you name.

What you do need is a replication of the storage backend. These do exist but it is a paid third party product that you have to find. THen you can have HA storage and build a HA cluster on the front end to fail over VM's between the Hyper-V host.

Without fgiving a specific product recommendation (what I name here is just one of many products and meant as an example) - you need something like http://www.kernsafe.com/product/hareplicator.aspx to simulate a high available storage system. It can not be done without other technology in the mix.

Or you use shared discs and a HA DAS raid controller - the only one I know is from LSI (https://www.lsi.com/company/newsroom/Pages/20120911pr.aspx).

But without something like this you just never will get proper data replication and high availability. Windows has no mechanisms for this out of the box.

  • LSI controller on this is link can do only cache synchronization you need to have disks still mounted into external SAS JBOD. There are two types of LSI Syncro: one CAN do what you want and other cannot. See: lsi.com/products/shared-das/pages/default.aspx So "8i" can be used to build "Cluster-in-a-Box" (no SAS JBOD) and "8e" cannot do it. Also these units are a) expensive ($1,000+ each) and b) you have single SAS path from every cluster node to every disk so config is a bit flaky. – BaronSamedi1958 Apr 5 '14 at 22:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.