[Serusers] [Sems] Poll on dropping unix socket interface ("unixsockctrl")

[Srinivas Kotamarti]

Stefan,

That's great. I should say the numbers are very impressive. Good work.

Certainly we will try to test our application with them and post the results.

Thanks

--Srinivas

-----Original Message-----
From: Stefan Sayer [mailto:stefan.sayer at iptego.com]
Sent: Wednesday, March 05, 2008 10:52 AM
To: Raphael Coeffic
Cc: Srinivas Kotamarti; serusers at lists.iptel.org; sems at lists.iptel.org
Subject: Re: [Sems] Poll on dropping unix socket interface ("unixsockctrl")

Hello,

Raphael Coeffic wrote:
> Hi Srinivas,
>
> this is a very good question you are bringing! I forgot to describe these
> aspects. So i will try to describe the experience we gathered in the last
> weeks.
>
> binrpcctrl:
>  - increases signaling performance (100% according to Stefan's measurements)
this is not correct, its much more:
http://lists.iptel.org/pipermail/semsdev/2008-January/002015.html
The relation is about 750 cps vs 80 cps, which is ~ 9 times performance
increase.


>  - increases flexibility (all the informations passed between SER and SEMS are
> configurable).
>
> sipctrl:
>  - increases signaling performance (80% against binrpcctrl according to
> Stefan's measurements)
I don't know where this figure comes from, it is as well much more (I
said this:
http://lists.iptel.org/pipermail/semsdev/2008-February/002221.html ,
1000 cps on lo on a laptop, versus a quad xeon over lan).

>  - reduces configuration and run-time complexity
>
> Concerning the stability, it is clear that the next weeks as well as the
> community will show us what those new control plug-in are worth. From my
> knowledge i would say that we are close to completion for both new plug-ins.
you can help by trying out your application with the new plugins and
report any possible issues.

>
> Dropping the support does not mean that we will remove it from SVN, that just
> means that future version should not use it. Existing installations are not
> concerned by those measures.
i.e. i would say it should be part of 1.0 release, but marked as
deprecated. After that release I would move for removing it, in favor of
more flexibility that new control plugins provide.

Stefan

>
> Cheers
> Raphael.
>
> On Wednesday 05 March 2008 15:08, Srinivas Kotamarti wrote:
>> Hi Raphael,
>> What are the advantages of binrpcctrl and sipctrl over the unixsockctrl?
>>
>> Has enough testing done on them? We are currently using SEMS with SER0.9.6
>> (with unixsockctrl) and it is working OK for us. IF we move to the
>> binrpcctrl or sipctrl would we get a better performance?
>>
>> Thanks
>>
>> --Srinivas
>>
>> -----Original Message-----
>> From: sems-bounces at lists.iptel.org [mailto:sems-bounces at lists.iptel.org] On
>> Behalf Of Raphael Coeffic Sent: Wednesday, March 05, 2008 7:00 AM
>> To: sems at lists.iptel.org
>> Cc: serusers at lists.iptel.org
>> Subject: [Sems] Poll on dropping unix socket interface ("unixsockctrl")
>>
>>
>> Hi *!
>>
>> with the emergence of the binary RPC interface for SER2 ("binrcpctrl"
>> plug-in), and SEMS' own sip stack ("sipctrl" plug-in), Stefan and I believe
>> that the unix socket control interface is defeated.
>>
>> That's why we would like to know who is in favor of dropping support for
>> the unix socket interface.
>>
>> Cheers and happy polling!
>>
>> Stefan & Raphael.
>> _______________________________________________
>> Sems mailing list
>> Sems at lists.iptel.org
>> http://lists.iptel.org/mailman/listinfo/sems
>

--
Stefan Sayer
VoIP Services

stefan.sayer at iptego.com
www.iptego.com

iptego GmbH
Am Borsigturm 40
13507 Berlin
Germany

Amtsgericht Charlottenburg, HRB 101010
Geschaeftsfuehrer: Alexander Hoffmann