First published on TechNet on Oct 01, 2013
As of September 2013, sixteen years prior, Microsoft released the Enterprise Edition of Windows NT Server 4.0 which included Microsoft Cluster Server (also known as MSCS). My how server clustering has changed over the years! In one of my last posts to Ask PFE Platforms as I plan to retire from Microsoft in October, this seemed like a good time to reflect on the sweet sixteen milestone for Server Clustering in Windows Server. My history with clustering goes back a bit in time and includes time well spent as one of the few beta engineers for MSCS prior to initial product release.
Early Days of Microsoft Clustering
Prior to joining Microsoft, I was a systems programmer in the mini/mainframe world and got some experience with clustering. In fact, in true geek fashion, I had my own cluster at home built out of surplus hardware and hobbyist licenses. I was a big fan of Dave Cutler and the emerging Windows NT operating system. When I jumped on board with Microsoft in March of 1995, I supported Windows NT 3.1 and 3.5…and a few other corporate level server products. I had no idea that just over a year later I would become a beta engineer for clustering using Windows NT…nor did I know that I’d still be involved with clustering many years later…or that I’d have a cluster in the corner of my basement keeping my home automation system highly available. When I first joined the project as a beta engineer in 1996, it was definitely an exciting time following the release of Windows NT Server 4.0. With prior clustering experience, I was always about design with redundancy in mind. This became quite the funny joke around the office because after joining the beta, my wife became pregnant with twins. Not only did the cluster beta guy have multiple two node clusters in his cube, he was now going to have redundant kids! As a precaution since product release was approaching, we found a volunteer to learn clustering and be my stand-in in case I became unavailable earlier than expected. Within two months, his wife was also pregnant with twins. The laughing stopped. Others in the office stayed way clear of our cubes (and the water cooler) until after product release. Was I ever glad that clustering at the time was limited to a maximum of two nodes! J When Microsoft Cluster Server released in September 1997 with Windows NT Server 4.0, Enterprise Edition, clustering began to provide high availability in many high visibility situations. I would find myself on the phone supporting clusters around the globe and meeting lots of great people. While MSCS was a released product for quite some time, people continued to refer to it by its beta name “Wolfpack”. Clustering in those days was limited to two servers coupled together as nodes and was very dependent on hardware that needed to be thoroughly tested, and on a certified list. Many issues were due to hardware, having the wrong drivers or firmware, or all of the above. When implemented correctly with proper hardware, etc…it worked great.
Evolution of Microsoft Clustering
Clustering improved with Windows 2000 but even more so with Windows Server 2003. Clusters could easily go beyond two nodes, use more memory, and utilize more processing power. In fact, there are still quite a number of Windows Server 2003 based clusters in existence today as the extended support end date for Windows Server 2003 approaches . Some of those clusters are geographically dispersed. However, Windows Server 2008 was a significant leap forward for clustering. Not only were clusters able to support even more nodes, memory, and processors…2008 x64 based clusters had support for Hyper-V virtualization built-in. Another fantastic improvement was the built-in validation process that alleviated the need for the old Hardware Compatibility List. Windows Server 2008 R2 took virtualization to another level with Cluster Shared Volumes and Live Migration .
Server Clustering Today
Windows Server 2012 expanded upon the great leaps forward of Windows Server 2008 R2 clustering by providing support for virtualization features like Hyper-V Replica . Windows Server 2012 clustering also scales to 64 nodes and up to 8000 virtual machines. That’s a pretty significant difference from the original limitation of only two nodes. For an example of how to create a Windows Server 2012 Failover Cluster, click here . Based on Brad Anderson’s post about Windows Server 2012 R2 release to manufacturing , general availability is right around the corner. In fact, his later post marks availability to TechNet and MSDN subscribers . If you are a subscriber, what are you waiting for? Downloaded it yet? Server Clustering once again continues to evolve and is easy to install. For a great summary of improvements and features, check out What’s New in Failover Clustering in Windows Server 2012 R2 on TechNet!
For More Information
Windows Server 2012 Failover Clustering http://technet.microsoft.com/en-us/library/hh831579.aspx
Windows Server 2008 R2 Failover Clustering http://technet.microsoft.com/en-us/library/ff182338(v=WS.10).aspx
Should I use the same network adapters for all interfaces on my cluster? http://blogs.technet.com/b/askpfeplat/archive/2013/02/21/mailbag-should-i-use-the-same-network-ada...
Clustering: What exactly is a File Share Witness and when should I use one? http://blogs.technet.com/b/askpfeplat/archive/2013/06/05/3506327.aspx
Cluster Team Blog on Technet http://blogs.msdn.com/b/clustering/
Brad Anderson’s ‘In the Cloud’ Blog http://blogs.technet.com/b/in_the_cloud/
Thanks so much for reading and rating my posts here...and for all the great feedback. I have certainly enjoyed being part of the community and interacting with those that have posted comments or needed additional follow-up.
Until we meet again!
Martin
