[erlang-questions] Matching IP address in socket module

[Loïc Hoguin]

Hello,

I would like to do a similar experiment with RabbitMQ. Would you mind 
sharing the code that makes 'socket' work similar to 'gen_tcp'? That may 
give me a head start, even if the code is not complete.

Thanks,

On 25/10/2019 12:39, Micael Karlberg wrote:
> Hi,
> 
> Its early days still, but the goal is definitely that it should
> be faster.
> 
> Here is some figures from a (time) test tool which is part of the
> test suite (its basically a ping-pong case), both server and client
> running on the same host (but in different VMs):
> 
> Two transports:
>          gen:  gen_tcp
>          sock: socket (tcp)
> 
> The socket active mode is a simulation of gen_tcp active mode
> (no active = N).
> 
> The tables are the exchange from the client side.
> 
> With server side using gen (active = false):
> 
>          Transport       Active          Data
>          gen             false           10192 byte/ms, 154 msgs/ms
>          gen             true            10383 byte/ms, 157 msgs/ms
>          gen             once            6003 byte/ms,  90  msgs/ms
>          sock            false           14050 byte/ms, 212 msgs/ms
>          sock            true            14772 byte/ms, 223 msgs/ms
>          sock            once            14050 byte/ms, 210 msgs/ms
> 
> 
> With server side using gen (active = true):
> 
>          Transport       Active          Data
>          gen             false           9447 byte/ms,  143 msgs/ms
>          gen             true            22345 byte/ms, 338 msgs/ms
>          gen             once            5532 byte/ms,  83  msgs/ms
>          sock            false           15316 byte/ms, 232 msgs/ms
>          sock            true            23693 byte/ms, 358 msgs/ms
>          sock            once            22068 byte/ms, 334 msgs/ms
> 
> 
> With server side using sock (active = false, async = true):
> 
>          Transport       Active          Data
>          gen             false           11260 byte/ms, 170 msgs/ms
>          gen             true            22273 byte/ms, 337 msgs/ms
>          gen             once            7703 byte/ms,  116 msgs/ms
>          sock            false           15211 byte/ms, 230 msgs/ms
>          sock            true            24778 byte/ms, 375 msgs/ms
>          sock            once            23086 byte/ms, 349 msgs/ms
> 
> 
> With server side using sock (active = true, async = true):
> 
>          Transport       Active          Data
>          gen             false           11351 byte/ms, 171 msgs/ms
>          gen             true            22469 byte/ms, 340 msgs/ms
>          gen             once            7407 byte/ms,  112 msgs/ms
>          sock            false           15484 byte/ms, 234 msgs/ms
>          sock            true            24885 byte/ms, 377 msgs/ms
>          sock            once            23570 byte/ms, 357 msgs/ms
> 
> 
> There is of course a bit of overhead since the socket transport
> is trying to emulate (part of) gen_tcp.
> 
> This is run on a Dell Precision T1700 running SLES 12 SP2 (not that
> that effects the relative performance).
> Run with a snapshot from the maint branch.
> 
> This is obviously not a real use case, but it can be a guideline.
> Also no UDP test (at the moment).
> 
> /BMK
> 
> On 2019-10-24 18:48, Frank Muller wrote:
>> Hi Thomas
>>
>> Is the socket module faster than gen_tcp/gen_udp in your case? If yes, 
>> can you please share some numbers.
>>
>> /Frank
>>
>>      > case socket:recvfrom(Sock, 0, nowait) of
>>      >       {ok, {#{family := Domain,
>>      >               port   := Port,
>>      >               addr   := Addr}, Data}} ->
>>      >       .
>>
>>     Hi Micael, Mark,
>>
>>     thanks for your replies. The snippet above helped me getting the 
>> record
>>     matched.
>>
>>     I'm quite impressed with the socket module, it seems to be pretty
>>     complete, at least for my application.
>>
>>     Thanks,
>>     Thomas
>>

-- 
Loïc Hoguin
https://ninenines.eu