Many things have been said about the complexity of today’s IT infrastructures, including that they are expensive and hard to manage. However, despite the many attempts to make things simpler, today’s IT infrastructure is more complex than ever. Remember how replacing the mainframe with PCs and UNIX servers was intended to make IT simpler? When someone tells you that things will soon be less complex, you should be skeptical.
In our search for the Holy Grail of simplicity, we seem to forget that many small steps often have the same (or better) result than a small number of large steps. For example, one of today’s bold moves is to look at cloud solutions to make IT simpler. But by now, many companies have already realized that cloud adds another layer of complexity to an already complex infrastructure. While cloud offers some strong benefits, it depends even more on components that we don’t have ownership of or fully control.
There are many things in IT we can do to be more efficient and save money. That’s exactly why the term “fit-for-purpose” makes so much sense. If you need a highly scalable, short-lived application that can accommodate anywhere between 100 and 100 million users, a cloud-based service will be the answer. In this case, you don’t need full control over all hardware; a simple Service-Level Agreement (SLA) on performance and uptime will be sufficient. However, if you need a new, mission-critical application that will live for years, accesses many existing applications and data points, has a predictable number of users and is economical, you’re probably better off running that in-house.
On the other hand, many of the applications that make money for us today are suffering from unreliable performance, unexpected downtime, an unacceptably long Time to Resolve (TTR), and SLA breaches that aren’t really helping keep the IT budget neutral. Isn’t it better to focus on fixing what makes money for us now instead of focusing on things that may make things easier in the future?
One of the reasons why IT is now an integral part of society is that we’ve invented and embraced many new technologies—and we must keep doing so. What’s different today is that we’re adding new technology on top of an already complex mix of technologies. The impact any new technology has on the existing stack is far less predictable than it has ever been. So, a little more focus on better managing what we already have will make implementing anything new a lot easier. In other words, let’s make sure we’re on solid ground before we jump into the clouds.
Managing What’s in Place
Most mission-critical applications today cross different platforms, application layers, and network topologies. Very few companies monitor these applications in an integrated way and this often results in unpredictable performance, a lot of finger-pointing when things go wrong, and high costs. If we aren’t in control today, changing these applications (i.e., replacing existing parts with cloud services, outsourcing parts of the processing, etc.) will make predictions about performance and stability very unreliable. Those companies that already do proper cross-platform management will find it much easier to replace existing components with a different technology or service. They know exactly what the impact of any component on the whole chain is and are in a much better position to make balanced decisions. They really kill two birds with one stone; they have better control over what’s making money for them today while they also are in a good position to make decisions about what will help them be more competitive in the future.
But simply monitoring x-platform applications takes time, and many mainframers say they no longer have time to either tune or fix performance issues. This is forcing them to do what they can to keep the mainframe running and only fix some of the most pressing problems. This situation has been caused by a lack of mainframers and the complexity of the mainframe itself. To be honest, in half the cases, the lack of people is caused by the complexity of the mainframe and the arcane tools we use to manage it. Solving a performance problem will involve products from different vendors all managing parts of the system. This includes VTAM, WebSphere MQ, the TCP/IP stack, applications in various languages, databases, the operating system itself, storage management, and many more components. It takes years of experience to master a subset of these well enough to find the cause of any problem.
So, why do expert mainframers spend so much time on mundane tasks such as installing software, applying maintenance, and other (easier) administrative duties? Shouldn’t they be working on complex performance and tuning tasks and, more important, transferring their knowledge to a younger generation? In the majority of IT shops, cost and staffing are the top issues. Less than a year ago, staffing was just one of many items on the priority list.
Is it possible to solve this problem? Yes, but maybe not in the short term; however, it can be solved by introducing a complete new way of interacting with the solutions you use to manage the mainframe. That workspace needs to use the power and knowledge of technology that has proved itself for more than 20 years and combine that with an interaction model that’s much more suited to manage today’s mainframe environments.
By introducing a role-based way of management instead of a product/solution-focused way, it looks a whole lot better while doing it. And that’s the danger when old veterans are presented with this kind of solution; they almost instantly respond with, “Seen it, done that, didn’t work.” But compared to the number of solutions used and the amount of information these veterans were provided, there’s no comparison. We manage the mainframe with fewer people today than ever and more and different products. Also, the information we deal with today (i.e., number of transactions, complexity, number of applications, connectivity, etc.) is much greater than say 20 years ago. It requires a different way of working. Your toolkit should help you with that.
So, with the availability of modern and integrated solutions that help you deal with the complexity of today’s IT systems, x-platform application monitoring doesn’t have to be hard. With the added benefit of managing a “business” process instead of a collection of IT components, it makes talking to the business managers in your company a whole lot easier as well.