[asterisk-users] VPN As SIP Tunneling?

Barry Fawthrop barry at ttienterprises.org
Mon Dec 11 16:48:58 MST 2006


Hi Anselm
Thanks for your input
Yes I was thinking of using OpenVPN so it was good to hear your experiences
I'm not so much concerned with the encryption of traffic etc..
But the Level of QoS.
If my IP Phone set QoS and the VoIP Termination provider's * PBX sets QoS

And we now connected via a VPN tunnel. We should be able to guarantee 
Quality due to the Tunnel.
The main issue is would I expect a higher latency ?
and (2) If I were using a 1 Mbps connect would I have less bandwidth due 
to overheads. That where I could do
8 concurrent calls x 115 bps 920 kbps  I could now only do 6  or will I 
still be able to do 8 ?

Thanks as always
Barry

Anselm Martin Hoffmeister wrote:
> Hi Barry,
>
> I used SIP over OpenVPN when travelling, especially from hotel rooms or
> showfloors. Of course I did not expect the performance of a local SIP
> connection, but generally it worked OK. The latency would not suffer
> much in comparison to direct connection, but a WLAN was involved which
> would screw quality anyway. Using a bluetooth headset would not help
> much, either. Having my geographic number from Europe ringing on my
> twinkle softphone over in California was nice-to-have
>
> Everything I say will be relevant to OpenVPN, which might be a bit
> different from IPsec, PPTP or other solutions.
>
> Am Montag, den 11.12.2006, 17:26 -0500 schrieb Barry Fawthrop:
>   
>> Hi All
>>
>> Could a VPN be used to help with SIP Tunneling and QoS issues.
>>
>> State 1:
>> Two IP Networks Connected via the Public Internet transmitting VoIP Traffic
>> Say a VoIP User and VoIP Termination Provider.
>> Each side can put QoS onto their part, but if QoS does NOT exist between 
>> them
>> then call quality will be bad anyhow.
>>
>> State 2:
>> Same as above except a VPN tunnel is setup between each side.
>> Thus making them appear on the same network and possibly same subnet.
>>
>> (1) Would this now traceroute a one hop ?
>>     
>
> Yep, but the packet containing the ICMP (ping) packet to traceroute the
> connection will itself be bumped around, so it will be the same number
> of hops really, just those between the VPN endpoints will be hidden.
>
> You win the bonus of getting around complicated NAT, possibly.
>
>   
>> (2) Would this have a lower or higher ping time, thus latency ?
>>     
>
> Higher, it can impossibly be faster than the packets carrying the VPN. I
> think you will not notice the difference though, because OpenVPN seems
> to do a good job.
>
>   
>> (3) With the additional Encryption etc.. if using a 1 Mbps Internet 
>> connection
>>     What would the "actual" amount available now be 700kbps, in other words
>>     How much overhead is there with a VPN tunnel that would reduce the 
>> available bandwidth ?
>>     
>
> Sorry, no numbers from me. For connection oriented protocols like HTTP,
> FTP, Mail, the additional problem of encapsulating TCP in TCP will kill
> the TCP windowing, as such not allowing for full line saturation (do not
> ask me for details, I slept to much during the networking lecture last
> year to tell you without a look into transcripts). No problem for UDP
> (like SIP/RTP). General data/overhead ratio will be probably better with
> larger data packets - I seem to remember you can configure the
> packetation size for some audio codecs in Asterisk. Alas I did not care,
> telephony was good enough.
>
> I would expect some throughput value wildly between 70 and 95%, but that
> is a guess, and not even an educated one. It probably depends on the VPN
> technology you use, and the maker(s) will be your authoritative data
> source there.
>
> Let us not forget that of course the data stream can be encrypted. Just
> that you think you are talking boring stuff does not mean there would be
> noone interested in wiretapping and listening in.
> (Even if I'm not paranoid they may be after me... ;>)
>
> Hth
> Anselm
>
> _______________________________________________
> --Bandwidth and Colocation provided by Easynews.com --
>
> asterisk-users mailing list
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users
>
>   


More information about the asterisk-users mailing list