OBiTALK Community

Region Specific Technical / Service Provider Support => North America - Including Google Voice, Skype, etc. => Topic started by: xpr722946ghd on May 13, 2017, 08:40:02 AM

Title: SIP Theory Questions re latency & SIP URIs
Post by: xpr722946ghd on May 13, 2017, 08:40:02 AM
OK, hopefully this a good forum to post my questions in.  I've been a long-time SIP user (over 12 years), but generally have been fairly close to the servers I register on.  Currently contemplating changing things around.

I live in Western Canada.  Some of the providers I might begin to use are located on East Coast of US or in Europe.  Ping times to their registration servers are 80-120ms for US, 200-220ms for European.  Typical ping to my nearest voip.ms server is 25ms.

Let's say I have a European DID and register to a European server on my OBI to receive the calls, I have that 205ms latency.

If I forward the European DID to a Voip.ms SIP URI, will I still have the 205ms latency, or is the calling party sent directly to my closer Voip.ms server?  Ways I could see it working.

1) Forwarding the SIP increases latency.  It has a hop to the Euro server, then a hop to the Voip.ms server.  Therefore 205ms+25ms = 230ms.

2) Forwarding decreases latency.  Upon arriving at POP, all traffic is redirected to the Voip.ms server bypassing the Euro server and going direct to my Voip.ms feed.

I could forward directly to OBI I'm sure and reduce latency, but I am wanting to make use of the voip.ms IVR systems on these external DIDs.

Perhaps there are other options I haven't thought of and neither of my theories are correct.
If someone could shed some light, or point me in the direction of a helpful article, I'd be very grateful.

Thanks!
Rob
Title: Re: SIP Theory Questions re latency & SIP URIs
Post by: drgeoff on May 13, 2017, 11:17:36 AM
End to end delay in VoIP has 4 constituents:

1. Codec delay. Insignificant for G. 711.

2. Packetisation at sending end and depacketisation at receiving end.

3. Transmission delay. Actual speed is somewhat less than the speed of light.  Even assuming 3x10^6 m/sec (186,000 miles/sec), crossing the Atlantic by cable takes about 20 msec.

4. Buffer delay in network switches along the route.

The ping times you measure are there and back but even so, indicate that #4 is the significant contributor. So the more hops you make the more the delay. Have a look at the times in a traceroute.

I would not worry about it. Any delay minimisations you can achieve will be minor and even without them I doubt your calls to Europe would be noticeably worse.
Title: Re: SIP Theory Questions re latency & SIP URIs
Post by: xpr722946ghd on May 13, 2017, 03:46:01 PM
Thanks for the explanation of the 4 areas.

So which is most efficient for the SIP journey?

Registering to a server in Amsterdam with a 220ms ping, or having that same server forward to a Sip URI at Voip.ms that is only a 23ms round-trip and just registering to that?

Or are you saying there is no effective difference between the two?  If that is the case, I'll just redirect to the voip.ms system where I can have the SIP calls enter my IVR.

From my home location there are
17 hops to Localphone in Amsterdam, taking 203ms.
14 hops to Anveo US taking 81ms.
15 hops to Anveo Canada taking 91ms
12 hops to TeleCallMart taking 103ms.
10 hops to my closest Voip.ms server at 25ms.

Goal is to have extra DIDs hitting an IVR.  Localphone is cheapest for incoming DIDs, TeleCallMart offer IVR system.  I don't mind using the Voip.ms IVR, or registering the Obi to TeleCallMart and using its IVR.

The main question is do I save any time / quality forwarding to a SIP URI from Localphone or TeleCallmart to Voip.ms, or is the higher quality / lowest latency achieved by registering the OBI directly to one of the aforementioned services?