Establishing what the estate actually looks like, not what the diagram says

Overview

A streaming provider had accumulated dozens of MySQL instances across teams with no consistent standard for replication, backup, or configuration. Nobody could state with confidence which databases could be recovered and how quickly. AceMQ was engaged to establish the real picture.

Challenge

Organic growth had produced meaningful configuration drift: instances with different binary log settings, replicas configured without GTIDs, and backup jobs that reported success while producing artifacts nobody had ever restored. Assessing this required verifying claims rather than collecting them.

Environment

MySQL estate on AWS spanning multiple teams and account boundaries, supporting content catalog and entitlement services.

Approach

AceMQ inventoried every instance and its actual runtime configuration rather than its intended configuration, then tested the parts that mattered. Backups were restored and timed. Failover paths were traced to see whether they terminated anywhere useful.

Solution

• Built a full inventory of instances with runtime configuration captured from the servers themselves
• Identified configuration drift in binary logging, GTID mode, durability settings, and character sets
• Performed timed test restores from existing backups to establish real recovery times and expose unusable artifacts
• Traced replication topology and failover paths to find replicas that could not actually be promoted
• Reviewed connection routing and proxy configuration for single points of failure
• Delivered a standards baseline plus a prioritized plan to bring drifted instances into line

Outcome

Technologies