...

1. Add a new headers length and value (byte[]) to the core message format.

2. Add a new Headers class

   1. It contains a map of key, value headers (
   Map<int, byte[]>) field to ProducerRecord and ConsumerRecord. 
3. Add accessor methods on the Producer/ConsumerRecord void setHeader(int, byte[]) and a byte[] getHeader(int)
4. Add ProduceRequest/ProduceResponse V4 which uses the new message format.
5. Add FetchRequest/FetchResponse V4 which uses the new message format.
6. The serialisation of the [int, bye[]] header set will on the wire using a strict format
7. Each headers value will be custom serialisable by the interceptors/plugins that use the header

For more detail information of the above changes, please refer to the Proposed Changes section.

Proposed Changes

There are four options proposed before this proposal. This details our proposed solution of Option 1 described here. The other options are in the Rejected Alternatives section.

Wire protocol change - use attribute bit4 as flag for if headers present. and add (optional) header size and field to the message format

The below is for the core Message wire protocol change needed to fit headers into the message.

A key to this is to ensure the headers cause no overhead if not present.

...

1. 1. map(Typed,Typed))
   2. It exposes accessor methods, which convert into typed values (this is akin to JMS/AMQP headers/property accessors)
      1. void setBoolean(String, Boolean)
      2. boolean getBoolean(String)
      3. void setByte(String, Byte)
      4. Byte getByte(String)
      5. void setChar(String, Char)
      6. Char getChar(String)
      7. void setShort(String, Short)
      8. Short getShort(String)
      9. void setInteger(String, Integer)
      10. Integer getInteger(String)
      11. void setLong(String, Long)
      12. Long getLong(String)
      13. void setFloat(String, Float)
      14. Float getFloat(String)
      15. void setDouble(String, Double)
      16. Double getDouble(String)
      17. void setString(String, String)
      18. String getString(String)
      19. void setBytes(String, byte[])
      20. byte[] getBytes(String)
   3. As noted Key interface is of type string, if this key though can be converted to a more compact typed representational form, for compactness it will be converted e.g. 
      1. "12" would be converted to Typed object of type=short and value=12 (int16)
      2. "true" would be converted to Typed object of type=boolean(true) and value=null
2. Add a new Typed class
   1. To hold the typed representations
3. Add accessor methods on the Producer/ConsumerRecord to get lazy initialised and lazy deserialized Headers Object.
   1. Headers getHeaders();
4. Add ProduceRequest/ProduceResponse V4 which uses the new message format.
5. Add FetchRequest/FetchResponse V4 which uses the new message format.
6. The serialisation of the [typed, typed] header set will on the wire using a strict format
7. Each headers value will be typed, where primitive types and string will be natively supported and support for custom types with a byte[] type.

For more detail information of the above changes, please refer to the Proposed Changes section.

Proposed Changes

There are four options proposed before this proposal. This details our proposed solution of Option 1 described here. The other options are in the Rejected Alternatives section.

Wire protocol change - use attribute bit4 as flag for if headers present. and add (optional) header size and field to the message format

The below is for the core Message wire protocol change needed to fit headers into the message.

A key to this is to ensure the headers cause no overhead if not present.

• The attributes flag bit is used to keep the message size the same as before if no headers are used
  
  

MessageAndOffset => Offset MessageSize Message
  Offset => int64  
  MessageSize

MessageAndOffset => Offset MessageSize Message
  Offset => int64  
  MessageSize => int32
  
  Message => Crc MagicByte Attributes Timestamp KeyLength Key HeadersLength Headers ValueLength Value
    Crc => int32
    MagicByte => int8  <---------------------- Bump up magic byte to 2
    Attributes => int8 <---------------------- Use Bit 4 as boolean flag for if headers present
    Timestamp => int64
    KeyLength => int32
  
  KeyMessage => bytes
Crc MagicByte Attributes  (optional)Timestamp KeyLength Key HeadersLength Headers ValueLength Value
    Crc => int32
    MagicByte => int8  <---------------- NEW [optional] size of the byte[] of the serialised headers if headers------ Bump up magic byte to 2
    (optional) HeadersAttributes => bytesint8 <---------------------- NEW [optional] serialised form of the headers Map<int, byte[]>Use Bit 4 as boolean flag for if headers present
    Timestamp => int64
    KeyLength ValueLength => int32
    ValueKey => bytes

Wire protocol of the headers bytes (if above mentioned attributes bit flag is true)

The below is for the headers wire protocol.

Headers (bytes) => Set(Key, ValueLength, Value)
  Set =>
	Key => byte [] => Typed <---    (optional) HeadersLength => int32 <------------------ NEW int[optional] keysize of the header
byte[] of the serialised Valueheaders => byte[]if headers
    (optional) Headers => Typedbytes <--------------------- NEW [optional] serialised form of the headers header value

Typed (bytes)Map<int, byte[]>
    ValueLength => (Type,int32
    Value)
	Type => bytes

Wire protocol of the headers bytes (if above mentioned attributes bit flag is true)

The below is for the headers wire protocol.

Headers (bytes) => Set(Key, ValueLength, Value)
  Set =>
	Key => byte [] => Typed <--------------------- NEW int key of the header
    Value => byte[] => Typed <-------------------- NEW serialised form of the header value

Typed (bytes) => (Type, Value)
	Type => byte <---------------------- 0x00 = null
										 0x01 = boolean=true
    byte <---------------------- 0x00 = null
										 0x01 = boolean=true
                                         0x02 = boolean=false
                                         0x03    0x02 = byteboolean=false
                                         0x040x03 = charbyte
                                         0x050x04 = short (int16)char
                                         0x060x05 = intshort (int32int16)
                                         0x070x06 = longint (int64int32)
                                         0x080x07 = floatlong (int64)
                                         0x090x08 = doublefloat
                                         0x0A0x09 = stringdouble
                                         0x0B0x0A = string
                                         0x0B = byte[]
    Value => bytes <-------------------- NEW byte array holding the corresponding value in byte array form, 
										 0x00-0x002 - null and boolean this will be zero length
                                         0x03-0x09 - type length, will be the length of the byte array
                                         0x0A-0x0B - string and byte this will contain the leading int32 length of the byte array.
 
 

Example Key Allocation

As mentioned above ranges of keys would will be reserved for different usages, below is an example way this could be managed.

Kafka open source  (0x000)

• 0x00000000-0x0000FFFF Kafka open source
• 0x00010000-0x000FFFFF RESERVED

Sample register that would end up having in the open source space.

To assit and help ensure ease of use and uniformity a constants class should be kept and updated with the above (similar to how java sql codes work) e.g.

package org.apache.kafka.common.config;

public class KafkaHeaderKeys
{

   public static final int CLIENT_ID_KEY = 1;
   public static final int CLUSTER_ID_KEY = 2;
   public static final int CORRELATION_ID_KEY = 3;
   
}

Sample register that would end up having local (In-house) custom per organisation

...

 type length, will be the length of the byte array
                                         0x0A-0x0B - string and byte this will contain the leading int32 length of the byte array.
 
 

...

Compatibility, Deprecation, and Migration Plan

• Current client users should not be affected, this is new api methods being added to client apis
• Message version allows clients expecting older message version, would not be impacted 
  • older producers simply would produce a message without headers
    • new consumer would simply get a message with empty headers
  • older consumers simply would consume a message oblivious to there being any headers
• Message version migration would be handled as like in KIP-32

...