On 09/05/14 16:06, Justin Cormack wrote:
>>   > [netmap, dpdk & snabb]
>>
>> Running the networking stack "behind" the kernel has slightly different
>> characteristics over a purely userspace stack:
>>
>> * applications use the kernel-offered service by default
>> * syscall RPC is fast (let's not deny that it's RPC ;) so if you need >1
>> application to use the same networking stack, it might end up being a
>> better performing approach
>>
>
> Sure, was just looking at all the options. Some prototyping to see what
> performs well would be useful.

My point was that performance isn't the only item to look at ;)

If we threadjack a bit and talk strictly about userspace networking, I 
still postulate that the choice of packet I/O framework will not matter 
very much.  Packet I/O cost is around 50 cycles per batch processed 
packet, while turning data into a packet (and vice versa) is >2000 cycles.

So, I think that the best choice for most users will be the framework 
that performs best in other metrics such as ease-of-installation, 
hardware support, "ecosystem" (buzzword warning!) etc.

------------------------------------------------------------------------------
Is your legacy SCM system holding you back? Join Perforce May 7 to find out:
• 3 signs your SCM is hindering your productivity
• Requirements for releasing software faster
• Expert tips and advice for migrating your SCM now
http://p.sf.net/sfu/perforce
_______________________________________________
rumpkernel-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/rumpkernel-users

Reply via email to