Solidatus has four thematic areas where we believe it is critical to understand the flow of data. One of these is transformation and it is this area that has become a source of some debate recently. There is a lot of fashion about transformation currently with a large part of the debate driving the conversation around DevOps. Now, I like DevOps in principle, it is attractive to any software engineer who wants to get value out for a customer quickly with little risk.
Any problems, so the theory goes, can be sorted out by rolling back and your Canary customer will only notice if they are unlucky. Canaries are a good idea and hark back to the days of miners who took birds in cages down their mines to forewarn gas build up (the canaries dies first).
In real life, things are not as clear cut as the evangelists might have you believe. Sometimes there is a change that cannot be rolled out to a canary group of users, principally because it is one on such a large scale that it involves lots of groups of people. These more often than not changes that involve the evolution of a data model and a change to its environment. I’ve witnessed several such unavoidably systemic changes in operation in my career and I can tell you that they are scary beasts.
Having no insider information about what has been happening recently at TSB I can only talk in the abstract about my experience.
Technical debt begets risk. Many banks have systems in place that were put together in the 1970’s - Yes, you read that right, the 70’s. Changing these beasts are horror stories that would fascinate Sherlock Holmes, if he were in IT. You don’t often find documentation for these systems or the original developers to hand. Perhaps if you are lucky there might be some source code, more often than not this is missing too.
The spaghetti that inhabits these systems is mind boggling and, although the need to modernise is all too clear, the risk of touching them makes kicking the can down the road a simple choice.
If we are faced with an unavoidable change, one which means tearing down decades old architecture in one go, then there are a couple of realities that you need to face:
One: it is going to have problems - I mean fact facts, nothing is perfect and if you have to fix forward then things might be a mess for a while.
Two: if you don’t take the time to prepare, when the problems happen they will be worse.
Supportive management is key here, if they are forcing the pace, then they should be around to cushion the teams from pressure. This is where responsible management earns its money though. A good manager will understand the risk, they will have information at their fingertips and they will have confidence in their information. A good team will back up their manager with accurate, up-to-date information and give them advice on the level of risk. The risk appetite is down to the manager, but they must be operating from a position of power.
At this point I am going to get a little preachy. The manager is part of the team, they are the figure-head for sure, but they are part of the problem. I have worked with people who understand this, and I have worked with people who will throw away others to protect themselves. A good manager will take advice and then stand with their team.
The question then is how to get good advice in a form that can be evidenced and one that everyone is on the same page about. Transformation is all about planning and better planning reduced risk (remember reality one - problems will happen). Mapping out your enterprise data and metadata model, overlaying your confidence about your areas of change in a proven way of providing an operational language for your organisation. Then plan the change up front, prove to yourself that you have done all you can to reduce risk and share this with management. Management should insist on nothing less. Then, when you have problems, you can isolate issues, fix them and move on.
Solidatus is proven to provide that vital picture of the organisation, focusing on the change and helping to minimise difficulties. So, whether it is DevOp in a fully shift-left automated super organisation or it is waterfall through necessity, we can help.
Remember, change happens - prepare for it, don’t fail at it.
Like this blog? You may also like our blog 'Fail to prepare; Prepare to Fail: A GDPR Conundrum'.