Patch automation from 34% to 92%, and $4.2M a year that stopped being spent
The challenge
A Fortune 100 bank ran 18,400 database instances across four regions. Each quarterly cumulative update consumed eleven weeks of elapsed time, most of it spent agreeing a sequence that preserved availability-group quorum and negotiating windows with 62 application owners. Automation coverage sat at 34%, and the remaining two thirds absorbed most of the database engineering team’s capacity.
What was done
Discovery mapped the estate and its dependencies in the first nine days, surfacing 1,340 instances absent from the CMDB. Patch waves were then generated from the dependency graph rather than negotiated in a meeting, with quorum and replication constraints encoded as policy. Health gates between waves decided whether execution continued, failures re-queued into a trailing wave automatically, and the change record closed itself.
The difference was not speed of execution — it was that nobody had to negotiate a sequence any more.