Most of you are undoubtedly familiar with one of the biggest buzz terms in the industry by now: blockchain. If not, have a quick read of a great article from a few months ago in The Wall Street Journal, CIO Explainer: What Is Blockchain?
Blockchain started out as the basis for Bitcoin—the sexy but not-ready-for-prime-time cyber currency that has had more than its share of setbacks (remember the Mt. Gox disaster?). However, the underlying concepts, techniques and technologies are sound.
Blockchain going mainstream
Well, okay, it’s not yet actually mainstream, but currently many banks and financial services organizations are experimenting with blockchain testing: ING Bank, HSBC, UBS, UniCredit, and Lloyd’s of London, among them. They’re running various programs and, for the most part, they are using “new technology” platforms for these purposes. The reality today is that large organizations run their businesses on a variety of platforms – mainframe systems, commodity servers and smartphones. I don’t expect this to change just because of blockchain adoption. – more on that later. There are three types of implementation that I can see happening: private blockchain implementations, public implementations, and hybrid implementations that are mostly private but connect to the outside.
Private blockchain implementations
These are highly secure as all nodes in a network are implicitly trusted by all other nodes. There is little to no danger of network attacks on any of the nodes since all nodes are internal to a presumably secured environment. Some would say that a private or internal blockchain implementation is nothing more than just another database format. That may be true depending upon the implementation specifics, but at least it does introduce a measure of redundancy that isn’t there in the standard bank-clearing house model.
I suspect that in private blockchain implementations the notion of the blockchain will not necessarily be in a relationship model, but probably in an implementation where the information is organized more like a novel – where words relate to sentences, which relate to paragraphs, that are in pages, that belong to chapters, that are part of a volume in book 1 of a 3 part series …. Not unlike some existing technologies (like IMS) that have been around for a while.
Public blockchain implementations or private implementations that include outside nodes
What this means is multiple branches of a given bank connected to the various branches of other banks and other recognized market participants: brokers and other service providers. This is how blockchain will hit the big time, and how it will become an integral part of the core systems of the world’s biggest banks and financial services organizations. Now this isn’t the only way blockchain will be used: it will be used for Internet of Things (IoT), peer-to-peer data sharing, and possibly e-voting (hosted on the platform(s) that make most sense for those implementations) —but I believe that serious adoption by banks and financial services organizations will open the doors for full-scale market adoption of the technology.
I suspect that since the data associated with the large-scale financial services blockchain will be likely to relate to a system of record that is stored on the mainframe, it will make sense to connect this data closer to that system of record. Meanwhile, “the way” that you will actually use a blockchain will likely come through some type of distributed system like a mobile browser that will be closer to the system of engagement. Bear in mind that I am thinking through implementations here that are scaling at the size of current banking where literally thousands of people are placing blocks on the blockchain per second.
Blockchain and the Mainframe
Most blockchain activities going on right now are running on unreliable, low-utilization, high-TCO commodity servers. I believe that when blockchain goes mainstream—and what I mean by that is when the biggest banks and financial services organizations adopt it into their core systems—it will have to be either on the mainframe or associated with the mainframe. That’s because the mainframe is the system of record for the world’s biggest financial services organizations, and runs their transaction processing applications that are their most important core business differentiators.
For those under the impression that commodity servers are the cheapest alternative for most computing concerns, that is only true in small computing environments. Things change in large-scale transaction-intense environments like large banks, credit card companies, clearing houses, insurers, etc. In these extreme environments, commodity server implementations start costing considerably more than mainframe computing. (See Dr. Howard Rubin’s data.).
These banks all run mainframe systems today, and have no plans to change that for the foreseeable future. Their best bet would be to run their blockchain tests at least partially in their mainframe systems either on z/OS or on a Linux logical partition / LinuxONE. Why? Well, they’re running their high-volume transaction processing on their mainframe systems because they need to, and because the mainframe is extremely cost-effective when running large workloads and has unparalleled system capacity and transaction processing throughput. Not to mention the extra reliability, security, resiliency, etc.
The blockchain crossroads
When and if mainstream banks decide to go prime time with blockchain, and they start running all their transactions through it, they’ll have some big decisions to make. Do they gear up to build out a whole new hardware and software infrastructure based on commodity servers and reengineered / reverse engineered / totally new development transaction processing applications? (Anyone who thinks this is a trivial step is either fooling themselves, or is uninformed.)
Or do they leverage what they have now and what they’ve been investing in for the past few decades? Do they leverage their extremely valuable intellectual property, their high-value mainframe applications and the hardware that supports high levels of transaction processing better than any other platform on the planet?
Tough call for some. Many third-party vendors won’t even consider the obvious choice, not for one second. Banks and financial services organizations need to do the legwork to figure out what’s best and what’s not, and what the cost will be up front, and what the YOY costs are going to be. Getting only one or two of those concerns right has the real potential to cost an organization many, many millions every year. In their own Dollars, not Bitcoin BTCs.
In any careful study, the economics will point to the mainframe; blockchain development can proceed on any mainframe operating system, and with the toolsets that your leading-edge developers are accustomed to using.