WPC is Microsoft’s conference for partners. The delegates tend to be executives or account managers from the Microsoft partner community, and the content is not the usual technical level one should expect from an MMS or TechEd.
Yesterday Microsoft announced some Hyper-V features from Windows 8 (Windows Server 2012?). The first was that Hyper-V “3.0” will support “more than” 16 vCPUs per VM. That’s a nice add on for those larger VM’s, giving us 16+ simultaneous threads of execution. People are virtualising larger workloads as well as the usual/expected lighter ones because virtualisation offers solutions to more than just power/rack consolidation, e.g. fault tolerance. A bottleneck has been the ability to run larger multi-threaded workloads, and Windows Server “2012” Hyper-V will give us a potential solution for this.
One of the big reasons we adopt virtualisation is the ability to make DR (disaster recovery or business continuity) easier. Mid-to-enterprise businesses can afford really expensive SAN/WAN solutions for this. There’s a number of storage or backup replication solutions that can allow replication of virtual workloads over smaller lines for the small-medium enterprise (SME). Some are good, and some are downright rubbish to the point of being dangerous.
Microsoft is stepping in with a new feature called Hyper-V Replica. This will give us the ability to replicate VMs asynchronously. This means it will work over longer distances, with lower capacity lines, higher latency, and will be cheaper. It also means that there is a slight delay in replication of VM data. That’s unacceptable to a small set of the market who have regulatory/business needs for synchronous replication and will have to continue to look at those third party or expensive SAN/WAN replication solutions.
Thinking about Hyper-V Replica makes me wonder if there are other new features/upgrades that we haven’t been told about yet. This isn’t DFS-R as we know it. DFS-R requires a file to be closed before it can analyse it and replicate the change blocks. Maybe we have a new DFS-R but I’m sceptical of that. Maybe we have a “new” transactional file system? I say “new” because Microsoft has had a transactional file system for quite some time in the form of WinFS. This would allow the file system to track changes, and replicate them, all while keeping VMs in a consistent state in source and destination locations. Consistency is one of those things that has worried me in third party software based replication of VMs because they are unaware of things like in-VM database commits. Maybe a new WinFS could be aware? Potentially it could work in cluster-cluster replication (no mention of that in the reports I read this morning from WPC).
Good news: Hyper-V Replica will be a built-in feature with no extra charges, unlike something else we could mention 
I think that’s enough hot air and methane blown into the atmosphere for today.
Hi, I wish you could bring some more virtualization info from WPC:)
Just a side note: NTFS is already transaction based file system (or you may use it this way) have a look at http://msdn.microsoft.com/en-us/library/aa363764(VS.85).aspx
Libor
Simplicity in multi-site solutions is a big must. Awfully excited to try this Replica feature.
Me too. This is a beta I cannot wait to try out.