CDC isn’t fragile. But it’s not forgiving either.
There’s a particular kind of infrastructure problem that’s easy to underestimate: the one where everything appears to be working.
Change Data Capture (CDC) tends to sit in this category. It runs quietly in the background, data moves between systems, and as long as nobody’s asking questions, nobody’s worried. The risk is that when something does go wrong, a component drops, a failover doesn’t behave as expected, the gap between “fine” and “serious problem” can be very small.
That’s roughly where a Saudi Arabian bank found itself. They’d implemented CDC to replicate data from IBM VSAM to Kafka, and it was working. But the high availability design hadn’t kept pace with how critical that replication had become. If individual CDC components failed, there was no clean failover path. The solution was more brittle than the environment required.
One of Triton’s Senior Consultants, Robert Philo was brought in to address that.
The initial brief was to design and implement a proper HA solution. In practice, as is common with CDC work, the requirements evolved once the architecture was being looked at closely. The design ended up being expanded to support failover at the component level, not just the system as a whole, because that’s where the real operational risk was sitting.
There was also a longer-term consideration. The bank knew they’d eventually need to bring Oracle replication into the same environment. Rather than treating that as a future project to figure out later, Robert built flexibility into the architecture from the start. When the time comes, integration will be straightforward rather than a rebuild.
The result was a CDC setup that the bank could actually rely on, resilient, granular in how it handles failures, and built for where they’re heading rather than just where they are now.
A note on how CDC engagements tend to go
The scope expansion in this project isn’t unusual. CDC environments often look simple from the outside – replication from A to B – but once you’re in the details, the edge cases and dependencies start to surface. That’s not a sign something was missed; it’s just the nature of systems that have grown over time.
What it does mean is that the person working on it needs to be comfortable thinking architecturally, not just technically. Fixing the immediate issue without considering how the environment will evolve tends to create more work, not less.
Expert on Demand
Not sure how your CDC environment would hold up?
Robert is part of Triton’s Expert on Demand team, and right now, we’re offering 2 hours free with any of our experts, no commitment, no sales process.
If CDC replication is part of your critical infrastructure, it’s worth having someone who’s seen it at scale take a look. Most of the time, a fresh pair of eyes will either give you confidence that things are set up well or flag something worth addressing before it becomes a problem.
Meet our experts and claim your free 2 hours, you could be talking to one of our experts in a matter of days.
Get in touch:
Email: info@triton.co.uk
Call: +44(0) 870 2411 550