TCP stack lock contention with short-lived connections

[Julien Charbon]

On 20/05/15 16:57, Adrian Chadd wrote:
> On 20 May 2015 at 06:27, Julien Charbon <jch at freebsd.org> wrote:
>> On 26/05/14 15:36, Julien Charbon wrote:
>> [...]
>>
>>  For people interested about this short-lived TCP connection scalability
>> effort, you can subscribe to the review of our latest (and biggest so
>> far) change:
>>
>> Decompose TCP INP_INFO lock to increase short-lived connections scalability
>> https://reviews.freebsd.org/D2599
>>
>>  The main goal of this review is ideally to start the rough discussion
>> before BSDCan (and then discuss details in person at BSDCan), and ease
>> tests by other people with more exotic configurations (thanks Adrian for
>> your early tests).  This patch still improves the short-lived TCP
>> connection rate (setup and teardown) from 60k/sec to 150k/sec.
> 
> I'm using this in our testing lab at work. I get to around 105k/sec,
> but I think that's primarily because of other lock contention in the
> kernel (things doing unnecessary ioctl()s.)

 Nice, always good to see other people running/testing changes.  On pure
number size, I would add that benchmarks are basically lies [:)] as you
get highest numbers from ideal conditions that are never met in
production.  Then, more interesting(/less lie) part is the relative
improvement you get using the exact same setup and for us it is
currently from 60k/sec to 150k/sec.  As example, on our older
hardware/older configuration it was from 35k/sec to 105k/sec.

 My 2 cents.

--
Julien

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 496 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freebsd.org/pipermail/freebsd-net/attachments/20150521/7c3bc6e6/attachment.sig>