we've build a gRPC schema registry and want to see if other people are interested in this:
I know that most people just use git as a registry, and that's ok but I believe having a registry spec (and an implantation) can help the community in the following domains:
- having a defined gRPC schema API can help to have a common way to talk to gRPC/Proto registries. This can combat fragmentation on the API level
- having an API can help as a store from dynamically created descriptors. Example: we have a tool that data governance can use to define business contracts, the tool uses the API to interact. verify and store the descriptors.
- having a spec can help data (I live mainly in the data space, sorry) systems (like Kafka or Apache Beam) to get up-to-date descriptors for their data workload.
So I like to donate a proposal of grpc.registry.v1alpha1, as it is now. From there on interested people can start discussing.
The repo contains a working implementation of the spec that we're using. For us, it's the centrepiece of our big data solution. I would like to donate this to the gRPC ecosystem. It's been proven very useful, for the following features:
- Dynamic descriptors (created by tools) can be visualized in git (super traceability)
- pluggable backend gcs, file and memory (we're using Google Cloud Storage)
- Notification of changes (pluggable), we're using PubSub to restart our pipelines when descriptors changed
- our data pipelines talk to the registry to get up-to-data contracts
- our gRPC ingestion API's are generated on the fly (they use the proto-java implementation to make dynamic services)
OK, I've talked a lot of Big Data use cases but I'm sure API use cases are to be found as well.
As a reference, my slides of the gRPC conf talk of last year:
So, is this something that is interesting to the community on who can help me getting this integrated?