On New Year’s Eve, 2010, I stood on the Canadian Department of National Defence’s (DND) loading dock and took delivery of a brand new mainframe. It was a watershed moment. It had been more than seven years since we had hosted any mainframe equipment in this building and since then we’ve bought three more new ones in another data centre. It certainly wasn't like the old days when the mainframe filled the entire room and sucked up enough power to dim the lights. These modern mainframes look like the Coke machine in the corner now. And they are designed from the ground up to help dramatically increase data center efficiency by significantly improving performance and reducing power, cooling costs, and floor space requirements with security and automated management and tracking of IT resources.
It has been quite the journey. Just three short years ago, mainframes were literally and figuratively being pushed out of DND. With the growth of the distributed world and a veritable glut of Windows admins, walls were built up against "those expensive mainframes," who responded by putting up walls of their own. Since mainframes never "broke," nobody ever complained. Eventually, the only people who knew they existed were those who supported them and those who counted the money—and only one of those groups was interested in keeping them.
With a visionary, dedicated mainframe team, this “end-of-life mainframe” situation was turned around to showcase the platform as a truly cost-effective, multi-platform enterprise hosting solution. With IBM mainframes running in a virtualized Linux environment and a zBX implementation, we have been able to reduce our IT administrative burden and lower both CPU and application licensing costs. This modern mainframe environment also has great potential to address the needs of a shared services data centre and all associated hosting challenges.
Looking back, our mainframe journey contains many valuable insights:
Dedicated Mainframe Team
Mainframes, or enterprise servers, can do many things that other platforms cannot do. But it’s necessary to have a dedicated team of people who can do the work to keep the system running. This allows the team lead to go to look at division, strategy, how to sell the platform to the user base, to clients, and make sure it is being pushed forward as a very viable box.
“Mainframe-Like” Isn’t Mainframe
In these times of austerity, we’re often told to find new ways to use old technology. We often have to make tradeoffs. Many companies will tell you that they have “mainframe-like” characteristics. That doesn’t mean mainframe. There are things mainframe platforms can do that are very unique. For example, if you are a very process-intensive organization like a bank, insurance company, a government department, any organization that does statistical analysis—you need the processing power that doesn’t exist in other boxes. Make sure you invest in the real thing. Keeping enterprise data on an enterprise class server is imperative. There's a reason the big banks and insurance companies run on IBM System z. Enterprise data is at its best on a high-end processing environment.
Be Willing to Play Outside Your Sandbox
Like many government departments, DND is divided into a number of organizations. The IT structure was similar. Three and a half years ago we had Windows/Unix/Linux support in one branch. Storage and backup was in another. Mainframe, including our own storage and backup, was in our branch. This structure left us with a number of cultural issues to overcome. People put up walls. There was resistance to the “independent” mainframe shop. There was a lack of engineering support for mainframes. There was a perception that everything at DND was Windows. We realized we’d be more successful playing nicely with others and moving outside of the comfort zone of our own walls. At the same time, the organization was transforming and helping tear down walls. I was given the direction to build an enterprise platform. First we determined what we were really trying to accomplish. For some of our aging workforce, change was difficult. Accordingly, our mission/vision focused on the following: The strength of the mainframe is its stability. An application you wrote 35 years ago will still run on today’s mainframes. Because change is challenging, we also focused on the fact that this was the desired platform of choice, a key hosting platform and the stability of the platform was one reason for its success for 45 years. This mission still applies today with the emergence of Shared Services Canada. Change is disruptive. But from our experiences, we know the sooner we accept the change and move towards it the better off we will be in a shared environment.
Creativity is the Best Way to Face Complexity
We were now evolving into an environment that could take risks. By moving one-third of a group of applications to Linux on IBM System z10, we could easily save at least $1.5 million per year. A “perfect storm” between cutting costs, space requirements, cooling and rapid growth of the data centre was coming to a head. Our data centre, like all of them, was getting bigger and we wanted to use Linux applications to manage the growth. There were 4,200 proprietary applications inventoried in DND alone at the time. The idea of re-platforming consolidation and the use of virtualization came to the forefront, gaining more importance within the department, as storage requirements continued to grow.
We decided storage is storage. It should be managed enterprisewide by a storage group with one person in charge. We eliminated our siloed approach to storage and then moved all Linux and Unix hosting to an enterprise model. We were now growing and this was a viable option for hosting.
It will remain important to continue to leverage our strengths, put hosting where it belongs best for the application, bring applications closer to the data, and continue to reduce overall licensing costs for applications.