// the find
confluentinc/schema-registry
Confluent Schema Registry for Kafka
The de facto schema registry for Kafka — stores versioned Avro, JSON Schema, and Protobuf schemas behind a REST API, and ships serializers/deserializers that transparently wire schema IDs into Kafka message bytes. Built by Confluent, which means it's battle-tested at scale but also means the interesting features live in their commercial tier. Anyone running Kafka with structured data who wants schema evolution guarantees and a central contract store needs something like this.
Compatibility enforcement is genuinely useful: BACKWARD, FORWARD, FULL, and NONE modes let you control exactly how schemas can evolve per-subject, not just globally. The wire format (5-byte magic + schema ID prefix) is simple and well-documented, so writing your own client in any language is straightforward. Field-level encryption with pluggable KMS backends (AWS, GCP, Azure, HashiCorp Vault) is a serious feature that most schema registry alternatives don't have. The subject name strategies (TopicNameStrategy, RecordNameStrategy, TopicRecordNameStrategy) give you real flexibility in how schemas are namespaced per topic.
The Confluent Community License is not Apache 2.0 — the client libs are Apache but the server itself cannot be used to offer a competing hosted schema registry service, which matters if you're building a platform product. High availability requires Kafka as its own storage backend, meaning this isn't a standalone service you can run without a Kafka cluster nearby. The split between open-source and Confluent Cloud features (data contracts, schema linking, stream governance) is poorly documented in the OSS README — you'll hit paywalled features mid-adoption. Maven-only build with multiple internal dependencies (common, rest-utils) makes local development setup more friction than it should be for what is ultimately a metadata service.