Case study · Caching · FinOps

Switching cache serialization to Protobuf

A reusable Spring Boot annotation moved cached objects from Java serialization to Protobuf 3 with zero downtime and over $1M in yearly savings.

Result: Zero downtime; over $1M a year saved.

Context

Dozens of Java services cached objects in a shared cache tier.

Default Java serialization made cached objects large, which inflated cache infrastructure costs across many services.

Built a reusable Spring Boot annotation so services could switch to Protobuf 3 one at a time, reading both formats during the transition.

Concurrency

before100%

after0%

How replacing a single global Redis lock with an append-only event log removed lock starvation and write races in a high-concurrency service.

Caching

before100%

after56%

Key-level memory profiling and a shared Redis client library cut cache memory 44%, peak utilization from 94% to 56%, and ended connection drops.

Databases

before1.0×

after0.4×

Finding slow SQL behind partner-facing APIs with OpenTelemetry tracing, then removing expensive joins, partitioning indexes and shrinking payloads.