Evangelizing Mainframe
Print Email

Where Does a Mainframe Fit in Cloud Architecture?

Before we jump into seriously discussing the mainframe and the cloud, let’s get a few things out of the way.

Yes, I’ve heard all of the comments about the mainframe being the original cloud. Yes, I know that there are many different types and definitions of clouds, but for purposes of this article they are irrelevant. Yes, I know that IBM has a very compelling Linux on z Systems value proposition, which is probably the best implementation that one could have in a cloud provider.

I’m talking about the legendary IBM mainframe running the z/OS workhorse that the vast majority of large financial, insurance, health care, government, retail and utility companies use today to run their enterprises.

I’m reminded of last year’s SHARE conference in Pittsburgh, where I was on a panel called, “How Does Cloud Computing Impact the Data Center?” In a standing-room-only crowd that asked very few questions, I had the impression that people came looking for an answer to The Big Question: What about mainframes in the cloud?

I felt honored to be among the panelists who had some very interesting things to say in terms of IBM’s offerings. As I listened to them, I formulated two aspects of how I think the mainframe running z/OS fits into this architecture:

  1. As a phenomenally scalable, secure data vault accessed by cloud-resident applications. This was a crazy concept just a few years ago. Who would ever seriously consider the network latency from a data-hungry application accessing data stores across anything other than a wide-area network? Well, I certainly didn’t see this coming, but I see it today. Many geographically dispersed organizations today connect their teams through a small number of highly capable data centers offering network latency that averages less than 15 milliseconds. Could you achieve better performance with the application co-located with the data sources? Of course you can, but this is not a requirement for many applications running today. Customers are already doing this in the real world: cashing in on their investment of the mainframe as a data server.
  2. As an on-premise I/O monster leveraging elastic cloud-based storage. I love that term, “I/O monster,” and there is no I/O monster like the new IBM z13 system. Pick your favorite statistic on how much data—I’m talking business data, not pictures of cats or videos—the mainframe serves, and it’s easy to see how archives can benefit greatly from cloud-based storage options. Most of this archived data is kept for regulatory compliance, so why have it taking up valuable space and management headaches on the same premises as the mainframe?

These are just two very realistic ways that I see how the z/OS mainframe fits into a cloud architecture. What other realistic ways do you see? And if you wished you could have seen that panel last year, there’s good news: it’s being repeated at SHARE in Seattle (session 16311 on Thursday at 11:15 a.m.). I hope to see you there and, if you do come, please ask a question.

Bryan Smith focuses on IBM mainframe solutions as the CTO at Rocket Software, a global software company that has developed mainframe tools and solutions for the world’s leading businesses for 25 years.

Posted: 2/19/2015 3:40:39 PM by Bryan Smith

Print Email

Join Now!