[asterisk-biz] Is ISP Blocking VoIP

[Sergey Kuznetsov]

Conrad Wood wrote:
> On Tue, 2006-01-31 at 21:54 -0500, Sergey Kuznetsov wrote:
>   
>> I see the only one way to do it.
>> To make simple test program which will be started on both endpoints.
>> Both boxes do the ntpdate right before the tests.
>> then one box sends and receives the number of rtp packets with 
>> increasing sequence numbers and stores the timestamps in the log
>> the same is done on the second box, it sends and receives from/to the 
>> first box and stores the sequence numbers and timestamps into the log.
>> then they exchanges by the logs, and third perl script compares the 
>> result and shows lost packets and latency for each packet in text or 
>> graphics.
>>     
>
> Actually you wouldn't need ntpdate, because you're only interested in
> the difference of time when packets arrive, not the exact time.
> It hence would be sufficient to put a simple
> milliseconds-since-start-of-program into each packet. The receiving end
> can derive the time it takes to travel from the amount of milliseconds
> past since the programs started and and the declared milliseconds in the
> packet. (Neglecting of course the RTT of the inital handshake, which
> would need to be compensated for)
>   
Yeah, We can use this method as well. Use gettimeofday()  tv_sec * 
1000000 + tv_usec delta since start of the program.
We canl encode this data in the first few bytes of RTP payload. No one 
will know that this is the timestamp
and not actual voice. Even is someone will play if, they won't hear the 
distortion.
> And a program like that would be tremendously useful to test QoS if it
> can simulate "other load" on different ports as well.
>
>   
Possibly doable, but depends on proxy scanners and id thy implement 
stateful inspection.



All the Best!
Sergey.