Status

Current state: Under Discussion

Discussion thread: here (<- link to https://mail-archives.apache.org/mod_mbox/flink-dev/)

JIRA: here (<- link to https://issues.apache.org/jira/browse/FLINK-XXXX)

Released: 1.17

Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).

Motivation

While restoring from a snapshot, StateBackend will resolve the schema compatibilty to decide where to go.

Currently, if users use a customized serializer to process data, they have to make sure that the schema compatibility in the old serializer (maybe in Flink library) meets the need as users want. Or else they have to modify TypeSerializerSnapshot#resolveSchemaCompatibility of the old serializer. There are no ways for users to specify the compatiblity with the old serializer in the new serializer. It also makes scheme evolution not supported in some scenarios.

For exmaple, if users want to implement a customized serializer which is compatible with ValueSerializer, user have to modify the compatibility logic in ValueSerializer.ValueSerializerSnapshot#resolveSchemaCompatibility. (ValueSerializer is final so that it also could not be extended). 

So reversing the direction of resolving schema compatibility could improve the usability of schema evolution.

Public Interfaces

Add an extra method in TypeSerializer.java

@PublicEvolving
public abstract class TypeSerializer<T> {
	// Check whether the serializer is compatible with the old one.
    public TypeSerializerSchemaCompatibility<T> resolveSchemaCompatibility(TypeSerializerSnapshot<T> typeSerializerSnapshot) {
		// Incompatible as default
		return TypeSerializerSchemaCompatibility.incompatible();
    }
}

Proposed Changes

Because Both TypeSerializer and TypeSerializerSnapshot are public interfaces, we have several steps to migrate the logic.

Step 1

Add an extra method in TypeSerializer.java as above.

Typeserializer#resolveSchemaCompatibility will return TYPE.INCOMPATIBLE default.

All places where use TypeSerializerSnapshot#resolveSchemaCompatibility to check the compatibility will call Typeserializer#resolveSchemaCompatibility firstly then call the original TypeSerializerSnapshot#resolveSchemaCompatibility.

If the result of Typeserializer#resolveSchemaCompatibility is INCOMPATIBLE, it will continue to use the original logic of TypeSerializerSnapshot#resolveSchemaCompatibility to check, else it will just use the resolving result of Typeserializer#resolveSchemaCompatibility.

Users could implement their own logic of resolving compatibility in their new serializer and it will be called firstly by flink.

If users haven't implement Typeserializer#resolveSchemaCompatibility, all behaviors are same as before. 

Step 2

It will confuse users there are two similar methods (TypeSerializerSnapshot#resolveSchemaCompatibility and Typeserializer#resolveSchemaCompatibility).

So we propose deprecating TypeSerializerSnapshot#resolveSchemaCompatibility for the long run.

In the step 2, we need to

1. Implement the method in all inner Typeserializers
2. Mark TypeSerializerSnapshot#resolveSchemaCompatibility as deprecated

Step 3

After several stable version, remove TypeSerializerSnapshot#resolveSchemaCompatibility and related implementation.

All places where use TypeSerializerSnapshot#resolveSchemaCompatibility to check the compatibility will only call Typeserializer#resolveSchemaCompatibility.

Compatibility, Deprecation, and Migration Plan

The deprecation and migration plan are listed above.

Test Plan

1. Existing UT of resolving compatibility could verify whether the old behavior could work as before.
2. Add extra UT to verify the correctness of customized serializer.
3. Add ITCase and e2e test case to mock the user behavior.

Rejected Alternatives

None.