In 2016 we transferred the application landscape used by Achmea Investment Management to the Mission Critical Cloud. We have made this journey many times, and we know that the unexpected is always a possibility.
Getting Back on Track after a Setback
In 2016 we transferred the application landscape used by Achmea Investment Management to the Mission Critical Cloud. We have made this journey many times, and we know that the unexpected is always a possibility. But this time it did not go at all as we had expected. That’s when you really discover the importance of working well together. We discussed the transition with Erwin Jager, Manager of Functional Management & Data Management at Achmea Investment Management. For him, this journey and its difficulties began in 2015, when he had just returned from a sabbatical to do a triathlon. What he didn’t know was that a new triathlon was waiting for him at work...
“Since 2010, Schuberg Philis has managed the application landscape. One critical process is the order processing, where traders use applications such as Simcorp Dimension, Bloomberg, and all kinds of spreadsheets to keep track of things. You’d recognize it from the TV news, where you see traders working in front of multiple screens each, and graphs everywhere. Our contract with Schuberg Philis was up for renewal at the end of 2015, so we were considering future directions. At the time, the application landscape was running on our own servers at Schuberg Philis, but we realized that periodically updating the hardware was old-fashioned. A transition to the cloud would give us more flexibility and be cheaper in the long term. At that point we also looked at other possible partners, but none of them could provide us with the same quality for the price we had agreed.
In August 2015, we decided to go to the cloud. Not everyone likes the sound of that, and we had a lot of explaining to do, to Achmea, to our customers, to the central bank (De Nederlandsche Bank, DNB), and the Netherlands Authority for the Financial Markets, to show that the Schuberg Philis private cloud is secure. We wanted to go live on 1 January, but when we started to construct a parallel landscape, we already encountered problems. Certain management applications proved unsuitable for the cloud. Schuberg Philis themselves were in the middle of an expansion from NL1 cloud to NL2, which meant that the production environment was ready, but the test environment was not. What’s more, we were moving into new offices on 14 March 2016, and we had to be ready for the Brexit referendum and its effects on the share market. In short, there were many variables.
So the migration actually took place on 7 March. The new virtual desktops for the traders went live and at first everything seemed to go well. But on 15 March, when the traders started working in our new office, we had a problem. First, the applications seemed to be running slowly. Then some traders couldn’t work at all, because they had to wait so long before a number they typed in appeared on the screen. At first they thought that this was the result of moving into new offices and, when the delays continued, they thought it was just a minor incident that Schuberg Philis would fix, as they always had. The result was that their complaints were not being heard, and that was what really hurt the good name of Schuberg Philis.
I made an internal presentation in which I compared the old server space with a second-hand car and the cloud with a Tesla. I posed the question: do we buy another second-hand car, or do we switch to a future-oriented form of transport and deal with the transition problems? They went for the Tesla. Schuberg Philis set to work analyzing the problems and scheduled 20 interviews with a cross-section of the organization. I told them, ‘Steel yourself, they’re mad as hornets.’ Luckily, it was not that bad. Our people were mainly pleased that someone was listening to their problems.
Then the engineers set to work on the log files to analyze why the applications were freezing. The tests for the migration had gone well, but the engineers found they had not been aware how many applications traders would have running live simultaneously. The new application landscape was not suitable for this and required design modifications. In the end it was August before everything was working properly. In retrospect, Schuberg Philis should not have built the test phase around an average user; they should have done a crash test with extreme users. And we – as the steering committee – should have been more on top of the problems. We’ve learned a lot in that respect.
If we had known what was in store, would we still have gone to the cloud? Absolutely. Sticking with the old situation would have spared us disruption in the short term, but the situation would only have deteriorated in the long term. Since Schuberg Philis always promises 100% uptime, they not only bore all the additional costs but also made a sizable cash contribution to our innovation budget. They take their responsibilities seriously, and we appreciate that. It feels as if we planned a week by the sea together and ended up sailing around the world. Yet neither of us blamed the other for anything, throughout the journey. We just wanted to solve the problem together. So yes, it was another triathlon, and because we have completed it together, we now know better what we mean to one another.”