Between last week's BUILD conference and this week's SNIA Storage Developer Conference, Microsoft made some significant announcements to the future versions of Microsoft Windows and Hyper-V (Windows Server 8 and Hyper-V 3.0 also referred to as Windows Next internally at Microsoft).
Looking at Windows, they have modified the UI again (I can already hear people saying "just when I figured out where everything is…") this time making it more like the Windows 7 phone OS which means it will be touch enabled, use the tile layout currently available with the phone OS and operate more like a tablet. They also added significantly more Powershell commands, more integration opportunities, a Microsoft App Store and an increased focus on cloud technologies through networking and scalability. As one person told me from the BUILD conference, "there was non-stop cheering from the developer community".
The most significant change in my opinion to Hyper-V 3.0 (and there will be A LOT) is the ability to run Hyper-V over CIFS or a networking protocol (technically SMB 2.2). This isn't new to the virtualization world as VMware has had support for the file-based NFS protocol for years as has XenServer which supports both CIFS and NFS. Why the wait for Microsoft? It likely had to do with performance and scalability concerns which Microsoft has now addressed. This means customers will be able to deploy Hyper-V VMs using NAS in addition to SAN protocols like Fibre Channel, iSCSI and Fibre Channel over Ethernet (FCOE). This should reduce the complexity of storage management with Hyper-V for some customers and will certainly reduce the cost associated with deploying Hyper-V VMs.
Other impressive changes planned in Hyper-V 3.0? Windows Server 8 will support up to 160 processors, processor cores or threads as well as up to 2TB of memory. With the increase in hardware performance, Hyper-V will also scale up with support for up to 32 virtual CPUs and 512GB of memory (a significant increase over the current limitation of 4 virtual CPUs and 8GB of memory). They will also now support up to 16TB for virtual hard drives. Hyper-V clusters will also be able to support up to 63 nodes and 4000 VMs in a cluster! Wow.
Other major changes include a new virtual switch for networking and new disaster recovery to expand how an administrator can protect Hyper-V virtual machines. Both features include APIs to allow partners to plug-in and add features to monitor and manage network traffic between VMs as well as simplify disaster recovery to easily replicate VMs between servers and storage (more on these later).
Add all of this along with the upcoming changes to System Center 2012 and Microsoft has greatly improved their private cloud, virtualization and management strategy.
So what's that sound behind you VMware? Don't look now but it's Microsoft quickly catching up to you!
Someone pointed out a mistake in a comment I made via a private discussion that I wanted to clarify. CIFS not performing as well as block technologies isn't accurate and while I know what I was thinking, I didn't articulate it very well. The issue isn't in regards to how well it performs compared to Fibre Channel when comparing performance results but how reliable it is and the impact this can have. Because CIFS is stateful, if the connection is reset than a reconnection to a remote server is likely necessary and data flow will be restarted. This can lead to large amounts of network overhead and can cause performance problems as a result (wait times). Also, with CIFS archictures you can design it poorly to span multiple locations or long distances opposed to block technologies that have limitations in distances helping to reduce the chance for user error. Thanks for pointing this out and for reading the blog!
Posted by: Virtual Winfrastructure | 09/22/2011 at 09:53 AM