[sdnog] IPFIX, sFlow collector

[Mehrdad Arshad Rad]

Hi Nishal,

Yes, sample rates were 1 in 20K and the packet per seconds was greater than
15Kpps (ipfix much more than sflow) as I remember.
Please ping me once you get a chance to try it.

-Mehrdad

On Fri, Mar 24, 2017 at 2:14 AM, Nishal Goburdhan <nishal at controlfreak.co.za
> wrote:

> On 23 Mar 2017, at 19:40, Mehrdad Arshad Rad wrote:
>
> Hi all,
>>
>> I would like to share with you a new IPFIX and sFlow collector which I
>> developed recently. If you would like to try it please send me an email
>> directly then I can help you with the installation or if you have any
>> questions. the main keys of this project are performance and scalability.
>> It tested w/ 3Tbps traffic, sampling rate 20K and Juniper devices
>> (MX,QFX) on an old dell server/Ubuntu 14.04 (usage12% CPU 200MB RAM)
>>
>> https://github.com/VerizonDigital/vflow
>>
>
>
> hi,
> that looks very interesting, thanks!
> bookmarked to test later.
>
> in particular, the performance statistics are impressive, and what has my
> interest.  i have two questions :-)
> * is that a sample rate of 1 in 20,000 packets/frames (which is what i am
> guessing) or a ipfix/sflow export rate of 20,000 pps?
> * what was your exported ipfix/sflow pps rate for the testing that you
> did  (assuming a 1/20,000 sample size, on iMix traffic)?
>
> i ask because my particular interest is testing this, for sflow exports
> from an ixp that’s in the 100gb/s mark, and i’m being lazy and hoping that
> i can extrapolate numbers for scaling hardware, from your testing. :-)
> (also the documentation that i read quickly, mentions scaling this in pps
> received)
>
> thanks in advance,
> —n.
>



-- 
*M*ehrdad Arshad Rad
*P*rincipal Software Engineer
https://www.linkedin.com/in/mehrdadrad
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sdnog.sd/pipermail/sdnog/attachments/20170326/e3375a6e/attachment.html>