JSON to Apache Avro Schema Generator — Free Online Tool
Convert JSON to an Apache Avro schema in JSON format with namespace, type, and fields definitions using AI. Supports null unions and nested records. Free, no sign-up.
JSON to Apache Avro Schema Generator — Free Online Tool
About JSON to Apache Avro Schema Generator — Free Online Tool
This tool converts any JSON object or array into an Apache Avro schema, inferring record names, field types, and nullable unions so you can immediately use the output in Kafka pipelines, Hadoop jobs, or any Avro-based data platform. Developers use it to eliminate hand-authoring verbose Avro JSON schemas and to ensure schema registry compatibility without trial-and-error cycles.
JSON to Avro Type Mapping
| JSON Type | Avro Type | Notes |
|---|---|---|
| string | "string" | UTF-8 encoded; maps directly to Avro string primitive |
| number (integer) | "long" | Avro has no generic number; integers default to long (64-bit) |
| number (float) | "double" | Floating-point numbers map to double for maximum precision |
| boolean | "boolean" | Direct 1-to-1 mapping |
| null | "null" | Standalone null or combined in a union type |
| object | {"type":"record",...} | Becomes a named Avro record with a fields array |
| array | {"type":"array",...} | items field holds the inferred element type |
| null + value | ["null", "<type>"] | Nullable fields become a union; null is always listed first per convention |
Avro vs. Alternative Schema Formats
| Feature | Avro | JSON Schema | Protobuf | Parquet |
|---|---|---|---|---|
| Schema language | JSON | JSON | Proto IDL | N/A (storage format) |
| Schema evolution | Full (reader/writer) | Limited | Field numbers required | Via Hive/Spark metadata |
| Binary encoding | Yes (compact) | No | Yes (very compact) | Yes (columnar) |
| Kafka ecosystem support | First-class (Confluent SR) | Partial | Yes | No |
| Nullable fields | Union with null | "nullable" keyword | optional keyword | All fields nullable |
| Self-describing binary | Yes (includes schema) | No | No | Yes |
| Human-readable schema | Yes | Yes | Yes | No |
Frequently Asked Questions
What is an Avro schema and how is it structured?
An Avro schema is a JSON document that defines a data type used for serialization. Top-level schemas for records include a type of "record", a name, an optional namespace, and a fields array where each field has a name and a type. Primitive types like string, int, long, float, double, boolean, bytes, and null are referenced as JSON strings, while complex types like records, arrays, maps, and unions are JSON objects.
Why does JSON null map to a union in Avro instead of a simple null type?
Avro is strongly typed and does not allow a field to hold multiple types unless explicitly declared as a union. When a JSON field can be null or a value, Avro represents this as ["null", "string"] (or the relevant type), meaning the field accepts either. By convention, null is listed first in the union so the default value of null is valid without extra configuration. Omitting the union and using just the value type would cause deserialization errors whenever a null appears.
How do I use a generated Avro schema with Confluent Schema Registry?
POST the generated schema JSON to the Schema Registry REST API at /subjects/{topic-name}-value/versions with Content-Type application/vnd.schemaregistry.v1+json and a body of {"schema": "<escaped-schema-string>"}. The registry returns a schema ID you embed in Kafka message headers. Confluent's Java, Python, and Go serializers handle this automatically when you configure schema.registry.url in your producer/consumer settings.
Can Avro schema handle nested JSON objects and arrays?
Yes. Nested JSON objects become nested Avro record types, each requiring a unique name within the schema. Arrays become Avro array types with an items field that holds the inferred element type, which can itself be a record, primitive, or another array. Deep nesting is fully supported, but each nested record must have a distinct name to avoid schema conflicts in the Avro parser.
What is the difference between Avro "int" and "long" and which should I use?
Avro int is a 32-bit signed integer (range -2,147,483,648 to 2,147,483,647) while long is 64-bit (range ±9.2 × 10^18). JSON has no distinction between integer sizes, so converters typically default to long to avoid overflow for large IDs or timestamps. If you know your values fit within 32 bits and want to reduce schema size and serialization overhead, you can manually change the generated type to int.
If jsondecode.com saved you time, share it with your team
Free forever. No ads. No sign-up. Help other developers find it.