[VoiceOps] Splitting SIP+RTP PCAP files
abalashov at evaristesys.com
Fri Jun 25 18:09:49 EDT 2010
My personal perspective is to emphasize the importance of good
programming fundamentals. I cannot emphasise how often I've seen code
written for fancy pancy hardware platforms/offboard processors that
fails basic CS. Linear lookups instead of hashes or trees, failure to
grasp memory fragmentation, unnecessarily heavyweight system calls,
that kind of thing. The fast hardware turns out to be necessary just
to run something otherwise so defective.
I'm not saying good algorithmic and coding practices will overcome the
limitations of COTS hardware when trying to capture and process off 10
Gb interfaces at wire speed or anything like that. But it is amazing
how far a little optimization will get you. I get the feeling a lot
of folks writing this stuff today learned to program in an era when
playing save-the-bytes with limited resources was no longer
fashionable or, from certain points of view, necessary, so they
approach system programming the way they do Java accounting apps.
On Jun 25, 2010, at 2:51 PM, Nathan Stratton <nathan at robotics.net>
> On Wed, 23 Jun 2010, Nicholas Sten wrote:
>> If you find yourself in that gray area where COTS hardware can't
>> save the
>> day anymore, but you're not looking to spend Empirix money, Endace
>> some really good cards on which to develop your own very robust
> Hardware capture cards really become critical as your traffic grows.
> With cards like Endance your able to do hardware level filtering and
> your software still gets the nice libpcap interface it has been using.
> As a bridge gap before you spend the money on hardware capture cards
> check out PF Ring.
> Nathan Stratton CTO, BlinkMind, Inc.
> nathan at robotics.net nathan at blinkmind.com
> http://www.robotics.net http://
> VoiceOps mailing list
> VoiceOps at voiceops.org
More information about the VoiceOps