It is not "our" protocol, it is protocol designed by exchange and we need to support it, as we can not change it :).
Simple binary messages, with binary encoded numbers, etc. No string parsing, no syntax, nothing like this, only bytes and offsets. Think about TCP header, for example.
JSON is very inefficient both in bytes (32 bit price is 4 bytes in binary and could be 7+ bytes as string, think "1299.99" for example) and CPU: to parse "1299.99" you need burn a lot of cycles, and if it is number of cents stored as native 4-byte number you need 3 shifts and 4 binary ors at most, if you need to change endianness, and in most cases it is simple memory copy of 4 bytes, 1-2 CPU cycle.
When you have binary protocol, you could skip fields which you are not interested in as simple as "offset = offset + <filed-size>" (where <filed-size> is compile-time constant!) and in JSON you need to parse whole thing anyway.
Difference between converting binary packet to internal data structure and parsing JSON with same data to same structure could be ten-fold easily, and you need to be very creative to parse JSON without additional memory allocations (it is possible, but code becomes very dirty and fragile), and memory allocation and/or deallocation costs a lot, both in GC languages and languages with manual memory management.
Curious if the binary protocol uses floating point or a fixed point representation? Or is floating point with its rounding issues sufficient for the protocol's needs?
No GP but familiar with these protocols.
They use fixed point extensively; I can't even thing of an exchange protocol which would use floating point since the rounding issues would cause unnecessary and expensive problems.
Prices are integer fields. When converted to a decimal format, prices are in fixed point format with 6 whole number places followed by 4 decimal digits. The maximum price in OUCH 4.2 is $199,999.9900 (decimal, 7735939C hex). When entering market orders for a cross, use the special price of $214,748.3647 (decimal, 7FFFFFFF hex).
For NASDAQ it seems to have been something around 430k / share... Buffett's BRK shares threatened to hit that limit a couple months ago: https://news.ycombinator.com/item?id=27044044
Most of them use decimal fixed point. Sometimes exponent (decimal, not binary one!) is fixed per-protocol, sometimes per-instrument and sometimes per-message, it depends on exchange.
JSON is very inefficient both in bytes (32 bit price is 4 bytes in binary and could be 7+ bytes as string, think "1299.99" for example) and CPU: to parse "1299.99" you need burn a lot of cycles, and if it is number of cents stored as native 4-byte number you need 3 shifts and 4 binary ors at most, if you need to change endianness, and in most cases it is simple memory copy of 4 bytes, 1-2 CPU cycle.
When you have binary protocol, you could skip fields which you are not interested in as simple as "offset = offset + <filed-size>" (where <filed-size> is compile-time constant!) and in JSON you need to parse whole thing anyway.
Difference between converting binary packet to internal data structure and parsing JSON with same data to same structure could be ten-fold easily, and you need to be very creative to parse JSON without additional memory allocations (it is possible, but code becomes very dirty and fragile), and memory allocation and/or deallocation costs a lot, both in GC languages and languages with manual memory management.