> Protobuf is ok but if you actually look at how the serializers work, it's just too complex for what it achieves.
Yeah. I do remember a lot of workloads at Google where most of the CPU time was spent serializing/deserializing protos.
I feel like most high throughput distributed systems eventually reach a point where some part of it is constrained by de/serialization.
Not much is faster than protobuf except for zero copy formats.
I feel like most high throughput distributed systems eventually reach a point where some part of it is constrained by de/serialization.
Not much is faster than protobuf except for zero copy formats.