The needs that lead to this protocol are:
- Encryption and Authentication to protect
- Compression should be used to reduce bandwidth
- Round-trip latency should not damage throughput
- Application-level message acknowledgement
Sequence and ack behavior (including sliding window, etc) is similar to TCP, but instead of bytes, messages are the base unit.
A writer with a window size of 50 events can send up to 50 unacked events before blocking. A reader can acknowledge the 'last event' received to support bulk acknowledgements.
Reliable, ordered byte transport is ensured by using TCP (or TLS on top), and this protocol aims to provide reliable, application-level, message transport.
Currently this is to be handled by TLS.
This entire protocol is built to be layered on top of TCP or TLS.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+---------------+---------------+-------------------------------+
| version(1) | frame type | payload ... |
+---------------------------------------------------------------+
| payload continued... |
+---------------------------------------------------------------+
- SENT FROM WRITER ONLY
- frame type value: ASCII 'D' aka byte value 0x44
data is a map of string:string pairs. This is analogous to a Hash in Ruby, a JSON map, etc, but only strings are supported at this time.
Payload:
- 32bit unsigned sequence number
- 32bit 'pair' count (how many key/value sequences follow)
- 32bit unsigned key length followed by that many bytes for the key
- 32bit unsigned value length followed by that many bytes for the value
- repeat key/value 'count' times.
Sequence number roll-over: If you receive a sequence number less than the previous value, this signals that the sequence number has rolled over.
- SENT FROM WRITER ONLY
- frame type value: ASCII 'J' aka byte value 0x4a
data is json encoded.
Payload:
- 32bit unsigned sequence number
- 32bit payload length (length in bytes of embedded json document)
- 'length' bytes of json payload
Sequence number roll-over: If you receive a sequence number less than the previous value, this signals that the sequence number has rolled over.
- SENT FROM READER ONLY
- frame type value: ASCII 'A' aka byte value 0x41
Payload:
- 32bit unsigned sequence number.
Bulk acks are supported. If you receive data frames in sequence order 1,2,3,4,5,6, you can send an ack for '6' and the writer will take this to mean you are acknowledging all data frames before and including '6'.
- SENT FROM WRITER ONLY
- frame type value: ASCII 'W' aka byte value 0x57
Payload:
- 32bit unsigned window size value in units of whole data frames.
This frame is used to tell the reader the maximum number of unacknowledged data frames the writer will send before blocking for acks.
- SENT FROM WRITER ONLY
- frame type value: ASCII 'C' aka byte value 0x43
Payload:
- 32bit unsigned payload length
- 'length' bytes of compressed payload
This frame type allows you to compress many frames into a single compressed envelope and is useful for efficiently compressing many small data frames.
The compressed payload MUST contain full frames only, not partial frames. The uncompressed payload MUST be a valid frame stream by itself. As an example, you could have 3 data frames compressed into a single 'compressed' frame type: 1D{k,v}{k,v}1D{k,v}{k,v}1D{k,v}{k,v} - when uncompressed, you should process the uncompressed payload as you would reading uncompressed frames from the network.
TODO(sissel): It's likely this model is suboptimal, instead choose to use whole-stream compression z_stream in zlib (Zlib::ZStream in ruby) might be preferable.