The Legacy Problem

The problem of legacy systems often finds its way into the press and gets many an organisation exercised, usually alongside a big issue - either perceived or real.

Reflecting on a conversation with a colleague before Christmas and I noticed this paragraph in an article I came across yesterday which made me think about;

"....a major roadblock to achieving truly game-changing digital transformation is the decades’ worth of legacy tech systems that are still being used to run their operations."

Of course nobody wants to rely on a business critical system which is so outdated it does nothing but cause problems, takes an army of people to do basic maintenance and/or costs a fortune to support. Then again in some cases an old legacy system may be operating perfectly well!

I can't think of a single technology leader that sets out to purposely create a legacy problem for themselves, inevitably it's inherited or evolves to that point through changes in scope, feature creep, funding constraints etc and becomes increasingly more difficult to do anything about.

Often the solutions seems to be to jump straight to a wholesale replacement but given the scale of some of these systems the resulting programmes can take years, cost more than you think and probably leave you with solving a new legacy problem not long after.

For many organisations this isn't an easy challenge to solve but it is one we can think about differently. When we talk about legacy are we clear on where the problem lies, do we mean;

• applications and services we've built ourselves, products we've bought and run on premise, hardware, connectivity, support contracts, or most likely a combination?

Being clear on what's included can help determine the extent of the issues and determine whether there are specific themes to address, is the problem:

• cost, security, support, resources / skills, licensing, business processes etc....?

We can then start to think about breaking the problem up and determining whether there are opportunities to alleviate the problems way whether that's:

• deploying automation, middleware (api layer, data virtualisation), moving the data, taking a different security approach (least privilege access/zero trust), start to move to a component based architecture etc....

There is of course one hurdle that needs to be overcome to resolve many a legacy issue and that's the type of money involved, it's often capital that's easier to come by and opex/admin that's in short supply which in the world of cloud services isn't that helpful.

None of this is new and is probably an approach may of us are taking, but the narrative around legacy just gathers up a multitude of problems under one umbrella, which from the outside is an easy to solve, as someone not working in tech recently said to me:

"surely this legacy stuff is easy, don't you just stick it in the cloud?"

If we can break some of the issues down into themes we might have more success at solving many smaller problems than one large scary one.