OpenVMS and Scalability
For OpenVMS users, scalability has always equated with clustering and with a maximum supported 96 nodes in a cluster. “OpenVMS is the most scalable operating system on the planet” - is that still true today? Yes and no.
The problem is the definition of the term cluster. It is relatively simple with OpenVMS: “It is a group of computers running the OpenVMS operating system”; they can be configured to be active/active or active/passive; in a highly scalable environment they tend to be active/active with load balancing. Using AWS (other Cloud environments are available) as an example, a virtualised cluster is defined as “… a logical grouping of tasks or services. If you are running tasks or services that use the EC2 launch type, a cluster is also a grouping of container instances. If you are using capacity providers, a cluster is also a logical grouping of capacity providers…”.
So, the term is the same, but the definition is different – groups of computers versus groups of tasks or services – and that makes it more difficult to compare scalability options but let us look at what we have. If you want to scale up with an OpenVMS system you manually add more active nodes to the cluster (and in this case size does not matter); however if you want to scale up using AWS you can use something called the Auto Scaling facility. In my previous article in this series I talked about flexibility within a virtualised environment and here is another example of that flexibility. When you set up your configuration with AWS you can create an Auto Scaling template and this template tells the virtualisation software what to do if your application reaches a threshold in resources. If you need more storage, it allocates more disk space until the cap is reached; if you need more network capacity it can add more connections; and if you need more processing power it can automatically add more VM instances.
As you can see it is not a like-for-like comparison when it comes to scalability on OpenVMS and virtualised software, and the situation gets more complicated when you delve a little lower with the two definitions. OpenVMS talks about computers, each of which can contain a number of cores; however virtualised instances are just that – instances; and one instance could use less than one physical core, so as I said not like for like.
Getting back to my original question – “is OpenVMS is the most scalable operating system on the planet?” – possibly, although there are many examples of high number multi-node Linux clusters; however it is not the most scalable environment on the planet; that accolade must now reside with virtualised environments. Is virtualised scaling right for your environment? My advice is to have an architect look at your environment before committing to virtualisation. In most cases this type of scaling will prove beneficial especially with its inherent flexibility, but it is always worth checking first.
You should also remember that even though virtualisation provides scalability, OpenVMS clusters will always be available, and we can cluster physical and virtual instances in the same cluster!
By the way, I have heard many a story about how OpenVMS can have more than 96 nodes, and yes it can in theory, however only 96 of them will be supported.