Windows Server 2012: Part 5.4—Cluster Shared Volumes

Slalom Consultant Derek Martin

Slalom Consultant Derek Martin is an accomplished Microsoft systems developer and integrator, experienced in developing and deploying SharePoint and CRM solutions, integrating line of business applications, and leveraging existing infrastructure investments.

Slalom consultant and accomplished Microsoft systems developer Derek Martin sheds light on Windows Server 2012 (WS12) through his insightful blog series focusing on his research within the technical preview documentation, personal experimentation with the product, and thoughts of how they can apply to the real world as soon as it is released to manufacturing (RTM).

Sorry about the delay in continuing my series. In the time I was away, I attended Microsoft TechEd (always an impressive show) and got wrapped up with two different clients. All good things but I got a bit distracted. Here is the next installment of the Windows Server 2012 goodness—cluster shared volumes (CSVs).

If you aren’t familiar with CSVs, see this excellent post on TechNet for the full scoop. The basic rendition is this: let’s say you have a cluster of machines and you want to have common storage among all of the instances. This common storage would be used for like SQL Clusters, file server clusters, etc. The problem with CSVs were many, but the main culprits were: they only worked with certain workloads, they were hard to configure, and often broke (typically around patch Tuesday when the head node rebooted).

In Windows 2012, one of the foundational underpinnings of all that is good is storage (hence the sub series on just this). In 2012, like 2008R2, we have the ability to create CSVs for the Hyper-V role. This means that all of my hosts can share a common storage LUN (or SMB3 but that’s another post!) and hosts only write to the blocks they are supposed to because they are smart (they know which Virtual Machines are being run in the cluster by which host). The obvious benefit is that when you need to live migrate from one host to another in the same cluster, all you do is transfer the memory and bam, you’re done. No more LUN remapping which I swear never really worked anyways. CSVs have also been extended now to the file server role, which brings in impressive additional capabilities. But they didn’t stop there.

CSVs in 2012 are even more impressive when you combine cluster aware patching as described earlier as well as drain stopping clusters, taking host nodes into a maintenance mode and particularly with regard to backups. I don’t fully understand the changes to the VSS writers within Windows to fully appreciate all that was done here but I do know that backups are now much smarter and less impactful on storage arrays when snapshots need to be taken.

Here is a list of other goodies that CSVs (along with clustering services in general) bring to the table in Windows 2012:

  1. Bitlocker support for the CSV
  2. 63 host nodes per cluster
  3. 4000 VMs per cluster
  4. Continuous availability for storage
  5. Parallel backups support
  6. Better incremental and snapshot support

Finally, CSVs can also take advantage of other general storage enhancements that are amazing as well:

  1. Data deduplication
  2. New file system which supports way bigger drives and more efficient storage (4KB sectors)
  3. Encrypted hard drive support (full hardware offloading of Bitlockering)
  4. Support for ODX (Offline Data Transfer) which, if you haven’t seen, checkout the Day 1 Keynote of TechEd North America to see them use ODX to transfer data at 10 GIGABYTES per second (not gigabits…GIGABYTES)
  5. Full support for the SMI-S protocol so that Windows can talk directly to storage arrays and carve up LUNs and more.  Very useful to System Center.

If you are a nerd like me, you will love CSVs and clustering in general in 2012. It is the key to a flexible cloud…well, one of them. Tune in next time where I’ll talk about something completely different maybe!

5 Responses to Windows Server 2012: Part 5.4—Cluster Shared Volumes

  1. Rolf says:

    I really like CSVs, but is this something that has been extended to MS SQL Server?

  2. slalomderek says:

    Not yet – still just basic file services, however, if you use Server 2012 to build a scale out file cluster and then use SQL 2012’s shared volume clustering feature, you can get the same effect (the share itself is part of a CSV).

  3. Zoltan says:

    Can I use two separated storages for the same cluster as both CSVs?
    (obviously all vhosts are connected with both storages, via FC)
    What about the witness disk in this case?

    • Derek says:

      Sure – any disks that are presented to the cluster can be CSV with very few exceptions. I have a cluster that does just that where five LUNs come from one array and five more LUNs come from another. They are all assigned to CSV and get mounted into the mount point folder on C:\ClusterStorage with no issue. Insofar as disk witness – still has to be a dedicated LUN/Disk. Great question!

      • Zoltan says:

        I can have just one witness disk per cluster, right? So I have to choose which storage I use for this. If so, in this case what happens if the storage with the witness disk goes down.
        Will the cluster remains up and running without the witness? I guess so.
        And I think the vm-s that are located on the other storage will be still running.
        But what if one of my host goes down as well? – they will they be able to “discuss” the situation without witness?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: