[zeromq-dev] Thoughts on Linux End User Summit

[Martin Sustrik]

Hi all,

I've just read an interesting summary from Linux end user summit:

http://lwn.net/Articles/361508/

The interesting aspect is that among the issues stressed by corporate 
Linux users the most dominant seem to be networking/latency issues. Some 
quotes follow:

"He [NYSE] would also like a non-disruptive way to measure latencies, 
especially in the network stack."

"Anthony had some requests too, beginning with support for TCP offload 
engines. What Morgan Stanley really needs, though, is shorter network 
latencies. Trades are dependent on getting orders in quickly in response 
to events, and latencies work against that goal."

"Specific problems include latency in the network stack and the ability 
of networking to make use of large numbers of CPUs. TCP, he [bank of 
America] says, is not scalable, but it wasn't clear where the problems 
are. One request that was clear was a means by which messages could be 
sent to multiple destinations with a single system call - something akin 
to the proposed sendmmsg() system call. He suggested that the time has 
come to move beyond POSIX interfaces."

The latter is the most interesting. The call to move beyond POSIX 
sockets is in line with what 0MQ/2.0 does, i.e. extend the socket API to 
handle issues that were originally thought to belong to middleware layer 
rather than networking layer. The obvious intent is to move 
functionality closer to the hardware with a prospect of better 
performance and lower latency.

The interesting part is that the problem was articulated at OS summit 
hinting that the common feeling is enterprise environment at the moment 
is that at least some of the functionality should be moved down to the 
kernel level.

Any thoughts on the issue anybody? Is there a section of messaging 
functionality that would benefit from being moved down to the kernel? 
Any experience with that kind of thing anybody?

Martin