I have spent the last 20 years absorbed by computers, computing, and software and I like to think that I have a broad range of skills and reasonably diverse experience. I have recently discovered however, that within my range of skills I have a huge and glaring gap. What is this gap, you ask? What have I so failed to delve into in my many happy years of work with computers? Well, despite the length and breadth of my career, I have never really worked with or, in all honesty, thought much about mainframes. While I knew mainframes where everywhere, until very recently I had no real visceral understanding of how huge a gap not knowing more about mainframes was. So, here are the basics to get us started on a journey to fill this gap in. While this all seems obvious when I think about it there is a simple truth that I have always been aware of, but didn't ever really stop to think about until recently. In a very real sense the majority of the world's systems of record reside on mainframes. Mainframes form the backbone of the modern system of commerce. They do it in data centres all around the world, they aren't flashy, they don't beg attention, they just quietly provide the compute power for our global finance, banking, health care, utilities, insurance and government systems. (Just think about that for a second.) Mainframes have been consistently evolving to meet increasing demands for speed, resilience and security for the better part of the last 70 years and at every step they have evolved to meet the world's needs and they have done this while remaining, for all intents and purposes, almost universally backwards compatible. In a world of distributed computing where software is scaling out rather than up; where blockchains and distributed databases are becoming common place; how has the mainframe retained it's hold on the worlds critical systems? It is tempting to site previous investment in the technology and the cost of conversion as the main factor in this, however, while this does play a role it isn't by any means, the whole story. People have been calling out mainframes as dinosaurs held over from a previous computing era for decades but here they are still forming the backbone of modern commerce. When I started to really look at the question, the answer to why build up instead of out, it isn't all that complicated. Mainframes in all their varied forms and roles have four things in common: 1. They are fast, support a massive number of concurrent operations, and scale seamlessly. Mainframes excel at providing sub-second response times to very large queries, and support massive numbers of concurrent operations. Other computer architectures can't match the sheer speed and power of the modern mainframe. While the benefits of vast arrays of processors and huge amounts of memory cannot be understated, additional power and versatility of the mainframe comes from it's ability to manage incredible amounts of I/O. Mainframe systems are designed to connect to large numbers of I/O devices that are specially designed to manage I/O related processing. This is critical as these systems routinely access petabytes of storage and offloading the management of I/O greatly reduces load on the core system. In addition, to being massively powerful, mainframes simplify data concurrency management and eliminate dependence on inter-node networks. Distributed systems need to address the constant challenge of ensuring that they are accessing and updating the most current version of a record especially in financial systems. In addition they need to include network transit speeds and communication negotiation in all of their response times. Mainframes, while providing high levels of redundancy in their own platforms, are secure enough, powerful enough and support enough concurrent transactions to make it possible to, even in the highest volume environment, use your single version system of record for real time transactions. In part, this is achieved by hardware level management of shared access to disk and disk level locking. 2. Resilience and availability is in mainframe DNA. Mainframes are designed for continual availability. Software, hardware and security changes can be performed without interrupting running applications. These massive, complex systems, are known to boast mean times to failures that span years or even decades. There is no monthly shutdown of a mainframe to apply security patches, there is no stopping a mainframe to upgrade RAM or add more disk, there is no rebooting the system when you install new software. Mainframes do this by constantly monitoring and auto-repairing their own hardware, supporting huge virtualisation environments, and by using complex, sophisticated clustering systems that protect workloads from interruption. 3. Mainframes are secure. They are secure for a variety or reasons but three of the biggest ones are i) their massive compute power supports encryption on a scale that can't be matched by commodity computers. Terra-byte size databases running on mainframes can effectively and efficiently be secured at the record level ii) they are fundamentally different that commodity computers in their hardware architecture, operating systems, and software design and these differences make them less prone to attack and finally iii) because they do not need to distribute their computing power across multiple nodes they can process much more without ever exposing data to the dangers of transit. 4. All of this has been achieved while maintaining almost universal backwards compatibility. You can reasonably expect software written in the 70's to run effectively on a mainframe built in 2020. This final point comes back to not only the protection of investment but it ensures that features that work today will work tomorrow. Beyond reducing rework on the applications executing in the mainframe environment, it means that you can functionally eliminate the down stream effects of hardware changes on the myriad distributed systems that rely on the core functions of the mainframe. Why is this important? This is important because any digital modernisation strategy that seeks to change or disrupt how the world does business needs to take the mainframe into account. In all reality there is no large scale digital change that takes place without the involvement of a mainframe at some level. This means that the more we understand about the mainframe, the more we can leverage their unique strengths to build a better digital future. Clearly there is much more to learn about mainframes and why they remain so essential to our modern world. I hope you join me for more in my "Digital Change from the Mainframe Up" Series. Have a great day,
Cynthia