No 5. The Challenges of Core Banking Replacement

As we highlighted in the Evolution of Banking, the technology landscape has evolved over time. Each change has brought new capabilities that were not possible with previous versions. At the same time, memory and storage capacity have become cheaper and much larger.

To give an idea of how dramatic this change has been, consider that 1TB of storage capacity (Disk) in 1957 would cost $9.2 billion, whereas today, it costs just $14. 1TB of memory in 1957 would have cost a massive $411 trillion; today, it costs just $2,100.

Consider also that Moore’s law has held true for over 50 years, meaning that computer processing power has doubled every two years: we have gone from 1000 transistors on a chip in 1970 to over 50 billion transistors on a chip in 2020.

It would be a natural conclusion to say that old mainframes are a key challenge. While many legacy manufacturers no longer build mainframes, IBM has done a great job maintaining them and improving their capacity and capabilities. They have also provided migration tools to update software to leverage their hardware. Today, there are still over 10,000 mainframes in operation used by some of the largest companies in the world. According to IBM, mainframes handle 68% of the world's IT production workloads but account for only 6% of the IT cost.

Some also associate mainframes with slow batch processing. However, IBM CICS (Customer Information Control System) has been around for decades and provides high-volume real-time processing. With built-in redundancy, these mainframes not only process billions of transactions daily but provide 99.999% availability, which amounts to just 5 minutes of downtime a year! Modern mainframes can run languages like C++ and Java and run software like servers. 

Whilst the finger of blame is often pointed at mainframes for banks' inability to upgrade their core banking solutions, the bigger challenges are elsewhere, and it is no wonder why some banks hold onto their mainframes for dear life.

However, there are many challenges when considering the replacement of core banking:

Skills availability

 Typically, legacy core banking solutions have been developed in older mainframe languages like Assembly or, most popularly, COBOL. These languages are no longer considered modern and aren’t attractive to most IT graduates. However, with decades of development and maintenance, the biggest challenge is that systems can contain millions of lines of code. It is the knowledge of this code and how to make changes safely without breaking functionality elsewhere that is most problematic. People who have worked and understand these systems have or are retiring, so knowledge leaks are the main challenge for sustaining these solutions.

Data migration

Replacing the software is only part of the challenge. Many of these systems were developed when memory and storage were expensive, so their use of data was minimised. This led to date fields being truncated, like just having 2 numbers for the year in a date field and dropping the two numbers for the century. This was the cause of much concern and work as we approached the year 2000. Today, we know that with more data, we can do more and provide greater flexibility within our systems. Hence, migrating data alone will not provide benefits, as such core banking needs to be redesigned to take advantage of better data. Today, AI is a great example of what is possible with better data.

Re-integration

The task of re-writing or replacing core banking would be much simpler if core banking were a single system that didn’t share data with other systems. However, new requirements had to be fulfilled over time, driven by new regulations, products, processes, channels, and changes in customer behaviour. Sometimes, these could be facilitated in the core banking solution, but often, they were developed outside and integrated with the core to utilise customer or product data. Large banks typically can have tens, if not hundreds, of systems around their core. So, replacing a core would need to ensure all these integrations can be fulfilled by the new system. Retesting all this alone is an enormous task, let alone the development effort.

Multiple cores

As I highlighted previously, when we looked at the evolution of banking, most banks have multiple core banking solutions. So, the bank would have the challenges described above multiple times over.

Legacy design

Perhaps the most understated challenge is redesigning core banking to provide greater flexibility and agility to make changes and drive innovation. With the benefit of hindsight of decades of evolution, any developer would design core banking solutions differently to leverage the benefits of the latest available technology. As we explored earlier, the evolution of technology has brought different benefits with each shift, and as such, every system should be developed with consideration for its eventual replacement. Alipay has developed its own core banking solution in China to serve over 900 million users. They have completely rewritten their core banking system four times in the last ten years!

These are some of the key challenges when considering a re-write or replacement of a legacy core banking solution. It should be noted that legacy need not be just mainframe / COBOL. There are client/server core banking solutions that also need to be replaced. Some core banking vendors have simply wrapped their software so that it can be run in the cloud, but this does not leverage the full capabilities of the cloud, which a true cloud-native solution provides. However, even cloud-native solutions are now outdated, as we will see when discussing the next generation of core banking software, Coreless banking.