"The mainframe will be dead in ten years at most" – I am sure you are familiar with these types of statements. What speaks against these bold proclamations is that they have been around for at least twenty years. First client/server systems were supposed to herald the end of the computing dinosaur, today it is the mainframe migration to the cloud. And there is definitely something truth in the latter.
Maybe this is because we fall into archaic thought patterns when the term "mainframe" is mentioned. We embark on a journey back in time to the 60s – or perhaps even further, back to the 40s, when Grace Hooper found the first bug in IT history. We see monsters weighing five tons and stretching 16 meters, like the tube-equipped Mark 1 computer (which also reminds us of a stately dinosaur in terms of its size). We feel the heat hitting our faces, we see the punch cards with the programs...
Without mainframe, IT would not be what it is today. In the meantime, of course, the tubes have disappeared. The latest generations of mainframes range in size from cabinet size to 19" rack units for cloud data centers and are indispensable in many areas. Did you know that mainframes handle 90 percent of all credit card transactions? Individual systems process billions of such transactions every day – reliably and securely. The modern world of online commerce would collapse without mainframes. Banks and insurance companies trust the giants. Still.
When I talk to insurance companies, the picture is clear across the board. Some develop new software, some use standard software, and some focus on developing new themes on modern platforms. A great example in this context is Allianz, which is clearly committed to a cloud strategy, including for its core insurance systems.
The future of mainframe applications is essentially dependent on strategic decisions – and these, in turn, must balance technological and financial opportunities and challenges. If it remains the first choice for customers, they will need to modernize workloads on the mainframe as well. However, if they decide to replace the mainframe, then migration to the cloud is a viable option.
The cloud not only lowers infrastructure costs, but also brings with it the option to use existing services such as AI functionality or automation for the applications. In this way, innovations for the mainframe applications can be quickly implemented. Microsoft's Azure platform, for example, offers more than 100 such services. A tried and tested approach is, for example, the replacement of DB2 databases with PostgreSQL and later with NoSQL databases such as Cosmos DB for certain applications.
However, the change in infrastructure (in the sense of a lift & shift) should only be the first step in a long-term modernization process. To replace the large cost items, it is necessary to free the applications from platform dependencies. This can lead to massive savings in license costs. Companies should be clear about the role mainframe computing should play in the future and which mainframe modernization strategy is best for them overall.