Impact of Serialization Format on Inter-Service Latency

Abstract

This study provides an evaluation of the impact of various serialization formats on inter-service communication performance, with a focus on serialization speed, space efficiency, and latency in environments integrating middleware, which are characteristics of microservice architectures. Through an empirical analysis of a wide range of serialization formats and comparison to the traditional standards, it highlights that the compactness of serialized payloads is more critical in reducing end-to-end latency than the sheer speed of serialization itself. Despite their high serialization speeds, FlatBuffers and Cap’n Proto underperform in distributed settings, in contrast to the more balanced performance seen with Avro, Thrift, and Protobuf. This study underscores the importance of message size optimization in boosting network efficiency and throughput.