Delivering faster with fewer incidents is something we continually aim for in technology. But experience has shown that achieving these aspirations is a lot more challenging when you are working with a vendor’s system rather than your own in-house one.
When we decided to accelerate the release rate for the HSBC investment bank’s global Transfer Agency (TA) system for transferring funds internationally we had to flex our thinking to work out how to increase the delivery cadence.
The challenge was that the TA runs on a core vendor system that supports 10 different jurisdictions in Asia and Europe and each one operates on a different regulatory framework.
The global maintenance period took place outside office hours for many of the jurisdictions. And every month, the vendor delivered more than 20 changes that had to be made at weekends, meaning team members having to work outside their normal hours.
And yet by the end of 2019 we had more than tripled the release rate from the same time a year earlier – how?
We decided to follow the slogan ‘Go Faster and Break Less’ to drive continuous improvement and understand what could be targeted.
Our first step was to look at the situation from a completely different perspective.
We analysed our historical releases and found that 60 per cent were data patches. That meant the changes were only being made to the database. Because we didn’t have to restart the application it meant there was no interruption to the service.
Since most data patches impact a single jurisdiction and each of their business hours are different, it was no longer necessary to follow the global maintenance timetable. Instead, we were able to schedule the releases more flexibly to meet local timelines.
Thirdly, we saw that most programme changes on the vendor’s systems were also only being made to the database. By utilising the deployment pipeline for programme changes we could automate 90 per cent of releases.
In the 12 months to December 2019 the number of incidents fell by 64 per cent and the rate of incidents logged as being ‘Caused by Change’ fell to less than two per cent of the total.
We standardised the post-release support procedure and documented detailed steps for everyone to follow in areas including data leakage prevention and risk controls. By automating 90 per cent of releases we also minimalised the number of manual deployments required which lessened the potential for errors.
We set up weekly incident review meetings to discuss individual incidents and service requests. If an incident ticket was created it would be closely tracked in order to accurately categorise what the issues were for future reference.
And better-support the Business
Since each of the 10 jurisdictions has its own priorities we now have a single point of contact in our team who faces off to their counterpart in the Business so their needs can be individually prioritised. We’re running a JIRA Service Desk with a simplified portal to raise and track tickets. That maximises the way we can deliver to them in a timely and transparent way which also strengthens our relationship with the Business.
We are already planning to roll out new features and functionality to build on the changes we delivered to the TA in 2019; we’ve shown we can go faster and break less and it’s given us the mandate to keep the momentum moving forward.