Bridging legacy systems and modern technology: The mainframe is dead, long live the mainframe
You rarely start from scratch when developing technology. There is always some legacy to take into account. In fact, the bridging of legacy systems and modern technology, is a business in itself. Never more so than in the world of mainframes. Although declared dead decades ago, they continue to run the world’s banking and insurance industry – and much more. Take a look at how two of the companies in the Data Respons group, Frobese and DONAT IT, are navigating between old and new.
Especially when it comes to Enterprise IT you might feel like an archeologist digging down into the past. The deeper you go the more fossils you find. But, although they’re ancient, these fossil systems are still an indispensable part of the infrastructure, connected to the upper layers by an intricate web of software and interdependency.
Although declared dead many times, mainframes are still around for a reason: Their ability to crunch numbers at speed is second to none, and they’re stable and reliable. You can replace hardware while the system is running, you can shift performance from one processor to another and always have full control over CPU and memory usage.
So, it should come as no surprise to you, that according to Patrick Leixner of DONAT IT, a German manufacturer of premium cars is generating and storing the “heart” of every new car in a mainframe: The secret keys for the car’s immobilizer.
Patrick is head of the DONAT IT mainframe team, currently consisting of 5 people, with two new colleagues joining soon. At DONAT, mainframes are a growing business area, including maintenance of the secret key storage system mentioned above. A niche, yes, but an expanding one, nevertheless. Mainly catering to the auto industry, Patrick and his team assists in maintaining mainframes running data warehouses and other essential systems.
For instance, a “just-in-time” system manages parts delivered by suppliers, collects data from the production lines and supply chains, consolidates it, and generates delivery notes, which are forwarded to the car manufacturer’s billing system.
Banking and insurance
Dirk Frobese, CEO of Frobese, specializing in software for the German banking and insurance sector, is clearly envious of Patrick Leixner’s team of mainframe experts. He himself is looking high and low for mainframe people.
In Frobeses line of business, mainframes are even more common than in automotive, but Dirk has only one employee and a few freelance developers with sufficient skills in mainframes and the programming language coming with them, Cobol. So, to meet the needs of his customers, he recently launched a new initiative called “´Kings of Cobol”, offering retired Cobol programmers lucrative freelance assignments. You’re fighting a losing battle if you try to get young engineers on board. They simply refuse to work with what the developer community considers “big iron”.
Patrick Leixner seems to be a notable exception to this. He’s by far the youngest member of the DONAT mainframe team. According to Patrick, he finds it fascinating to work with software distinctly older than himself.
More transactions than Google
The fact that mainframes are highly reliable number crunchers secures their survival in large enterprise IT infrastructures. Mainframes are used by 71 per cent of Fortune 500 companies. They handle 90 per cent of all credit card transactions, and host more daily transactions than Google. Airlines, banks, insurers, they all rely on mainframes.
But, however successful, not all is bliss and perfection in the world of Cobol and mainframes. Apart from the difficulties finding skilled people, according to Dirk, the main problem is Cobol CICS, the IBM transaction system which is used in the Cobol financial applications. To understand the context, let’s have a quick history lesson from Dirk:
- Nearly 30 years ago, I was fresh from university and started working with software for the banking and insurance sector. At that time, we had a vision to get rid of the host and replace it with client/server applications. Our customers wanted us to develop nicer user interfaces than the green-on-black screens they were used to. So, we began to develop Java applications to replace Cobol clients. Then came another tier, a server layer controlling the clients and running with Java. And then we made an API, a kind of middle frame, to transport the interaction from the user interface to the mainframe handling the transaction.
- In the end, everything looked good, but in the background, the system was still running all these Cobol CICS transactions. However, now management had no incentive to replace the host, because all the users were happy with this nice, modern user interface we had developed for them. This is when the trouble began.
Cobol CICS is the problem
As Dirk explains, the main problem is Cobol CICS. You can easily replace Cobol, but not the CICS. The CICS is the middleware, lying on top of the operating system and ensuring that a transaction is safe. In businesses like banking and insurance you have vast numbers of complicated transactions. If one of these transactions fails, for whatever reason, you can roll it back and find the cause of the failure. According to Dirk, there is no real replacement for the CICS, so migrating to a newer system involves a lot of risk.
In addition to this, mainframes are a monopolist business dominated by IBM as the sole vendor. As a customer, you pay for a license, without any alternative, and without any real competition between vendors. This means the price is rising slowly and steadily, and there’s nothing you can do.
Here to stay
That said, mainframes are here to stay, whether you like it or not. And, according to Dirk, they will still be part of the IT infrastructure for many years to come.
- Mainframe systems are ideal for number crunching and fast transactions. But for other kinds of banking and insurance logic, for instance calculation of loans, you can replace them with other platforms like Unix. So, just like when I started my career 30 years ago, we’ll have a heterogenous environment with small clients, middle tier servers etc., and a host at the end, running a big database, data warehouse or transaction machine. That concept is old, and it won’t change. We’re going to continue to see it in the future.
Finding skilled people
The crux of the matter is however, to find people willing and able to work with this “big iron” as it’s called in the business. Specialist companies like Frobese and DONAT IT are providing these skills for their customers, although Cobol programmers are few and far between. And both Frobese and DONAT are planning to expand this part of their business. Currently they are even discussing to join forces to beef up their crew of Cobol experts, because they predict a rising need for them.
In short: There is good busines in bridging new technology and legacy systems. The mainframe is dead, long live the mainframe!
The Frobese Kings of Cobol website (in German):