Articles
The Legacy Life-cycle
Which legacy system can you live with and which should you replace?
Carl Maughan ponders the question.
Legacy is back on the agenda. If it ever went away, that is.
A recent survey of banking professionals – an industry with one of the oldest proliferations of legacy technology – found that 76% of respondents believe their corporations will still be using the same legacy systems five years from now.
Some of these systems have been processing banking products for the last 40 years, operating live 24 hours a day, every day of the week. Their owners apparently see no convincing reason to replace them, or at least no reason good enough that it overpowers the fear of losing such valuable assets.
The window for procrastination is narrowing, however. The same survey, from independent researchers Spectrum Consultants, reported that over 90% of IT professionals, from 35 banks, say legacy systems are becoming more expensive to update and maintain. And that’s without considering the necessary integration of impending regulation.
What this actually reveals is that there are obviously some legacy systems you can leave for the moment and there are others you simply cannot afford to keep any longer. So which are they?
Clearly, many of those banks in the above survey are happy to leave their legacy systems quietly working away, processing transactions and managing accounts. These are typically referred to as ‘Cash Cows’.
Types of legacy
Cash Cows are the more pleasant face of legacy. You might not completely understand how they work - so they’re difficult to change - but they rarely need maintenance, they’re reliable, and they make you money. They have little negative impact on the business.
Then there are those legacy systems that can be regarded as ‘under maintenance’. It doesn’t take much to tip a Cash Cow into this area of legacy: a shift into a new area of business, or upgrades of your general IT infrastructure, anything that requires you to take the lid off your systems and tinker about inside. The important point here is that you have the skills and knowledge to ensure this maintenance isn’t a problem, although it does require a small amount of time and effort on the part of your IT department.
Turn this up a notch, and you’ve got a legacy system that’s starting to be a real pain. It requires almost constant maintenance, which also means the need for extensive testing to ensure clash-free code and reliable service from your legacy systems. Which can be expensive.
And it’s when you get into exactly that kind of situation, with regular code changes and ongoing maintenance, that there’s a chance your legacy system will slip into something far more unmanageable. Making even minor code changes to a system that’s hard to understand inevitably results in clashes elsewhere in the system, which results in endless testing and maintenance, which increases the risks that some other clash will arise. This, it has to be said, is a very expensive system that you need to replace as soon as possible.
Then, at the absolute end of the scale, is the Critical system that you need to replace immediately. It’s haemorrhaging money and is not fit for purpose; the original staff who built it are long gone, and your current IT department is spending all its time trying to understand how to keep it going – and failing.
It is worryingly easy for a Cash Cow system to become a Critical system. What trigger points could push it over the edge?
There are general trigger points as well as those specific to your industry: the software you’re using becomes ‘sunset’; your system is based on a burning platform; incoming legislation requires significant changes you can’t make; or mergers and acquisitions activity necessitates combining two or more legacy systems.
There could be any combination of events that require you to replace your legacy system, and you should recognise what they are before they happen. So what do you do?
Solutions
Thankfully, legacy replacement models have grown and changed almost as much as the technology itself, so there are quite a few options depending on your situation.
For instance, if you have a Cash Cow system or one that is consistently ‘under maintenance’, where you don’t really want to change anything about the system, you might just want to hand over day-to-day operations to someone else. Outsourcing the system to a hosting company is certainly an option here: it frees up your internal staff to look at other options like preparing for wholesale replacement.
But if you really need to replace those systems right now, then there’s a wider range of choices, not all of which carry the same level of risk, cost or effectiveness.
Total replacement might be your only option if your hardware is obsolete and the system source code so old and modified that you have little idea how it works. This approach is highly risky – you don’t know if some of your valuable processes will be lost in the transition or even whether you’ll get a replacement system with the same functionality.
There are dozens of solutions vying for business in this space, using a variety of technologies and methods to take your COBOL applications (or whatever they may be) and move them to more modern platforms. Some even make use of complex emulation to ensure your systems remain the same, just run on modern and supported technologies.
If you’re that risk-averse (and, let’s face it, if you’re dealing with millions of pounds worth of business then you very well could be) there’s always methods that allow you to modernise without opening up your precious legacy system. SOA, for instance, allows you to ‘wrap’ your applications with a more manageable layer, but it still leaves the potential legacy time-bomb lurking underneath.
More radical solutions also exist that can not only eliminate legacy but also much of the risk that comes with modernisation. Approaches exist that allow you to recover legacy system behaviour non-invasively (by recording the live operation of the system and not looking at source code) and then transfer this to an easily changeable platform to avoid legacy in the future.
Whatever method you choose, it should be the right one for you. Regular evaluation of you ‘legacy estate’ should be a matter of course for all businesses and, combined with foresight, should ensure that when a trigger point does appear, it doesn’t mean your business falls behind.