Description
Profiling Ruby Avro decoding of our event processing pipeline shows Avro is spending a fair amount of time validating that the reader and writer schemas are compatible. These checks are unnecessary for the fairly common case of the reader and writer schemas being the same. Eliminating these unnecessary checks when the reader and writer schemas are the same improves the throughput of our decoding benchmarks by 1.3X.
Github PR coming shortly.
Attachments
Issue Links
- links to