Voice over IP over FR?? 3

[renbd]

I've been told that there is a diference between Voice over IP, Voice over FR and Voice over IP over FR.
Now, does this make any sense? FR and IP are protocols from diferent levels, how can for example IP substitute FR?
Isn't IP (or any other network protocol) always needed for the voice packets to travel over a FR network? Or voice can reach its destiny only with the DLCI?
Or in other words, in a FR network isn't the voice, after it has been compressed, packed with the IP header before the FR header?

Thanks for any feedback!

 

[jeter]

You have to do traffic shaping for example and this is for VOFR !!!

map-class frame-relay voice
frame-relay voice bandwidth 48000
frame-relay fragment 360
no frame-relay adaptive-shaping
frame-relay cir 256000
frame-relay bc 2560
frame-relay mincir 64000
frame-relay fair-queue

when you set up your frame relay voice bandwidth add 12000 per each line to be used. This is how it maintains its connectivity . ( so I am told ) ! #-)

 

[wybnormal]

Also, dont forget to get a SLA that specificly says what the latency period will be. VOip hates latency and most frames if they are any distance runs from 20 to 120 mS. Also make sure you have not over committed the frame connection as many do

Mike S

 

[renbd]

excuse my inexperience but I didn't understand the diference between VoFR and VoIPoFR.
Maybe if I refrase my question: (well, first of all, I'll assume there is a diference between those two since you didn't say anything about that) where in the process is the diference (digitazing the voice to PCM, compressing, maybe header compressing, IP packeting if applies, FR packeting, ...)?

 

[majestik]

Rebd,


A VoIP solution is one where the end-station voice-node; an IP telephone or IP PBX, digitises the voice conversation using a sampling codec such as G711 or G729. It then creates an IP packet with a single piece of the codec data stream as payload and shoots it out onto your IP network, with Ethernet or ATM as the layer 2 transport protocol. A typical VoIPoEth packet will consist of:-

bytes
====
30 G729 codec 30ms of payload
02 H323 header
12 RTP header
08 UDP header
20 IP header
14 Ethernet header
04 802.1q header
06 Ethernet CRC checksum

So with only 30 bytes of payload, the IP packet is 72 bytes long and the Ethernet packet can be up to 96 bytes long if you are using 802.1q to prioritise traffic on your LAN. This just goes to show how inefficient VoIP is!

Anyway, once the VoIPoEth packet gets to your router, and assuming it is destined for a remote node over a Frame Relay WAN, the Ethernet information is stripped away and FR header information added. Assuming you are not using any form header compression, the new VoIPoFR packet would be 72+6=78 bytes.

VoFR is a different beast. TCP/IP doesn't come into the picture at all. You typically have an E&M card in your router which provides the quantisation process on analogue voice coming from a telephone or PBX tie line. Each segment of the codec data stream becomes one payload for a FR frame and the result is a much more efficient, reliable and stable form of telephony over data. You need to establish call-tables on the router so that it knows where to route the "call" based on dialled digit strings and DLCI paths. In a VoFR scenario, your roter becomes more like a PBX.


Regards,
Maj.

 

[renbd]

Well, it's been one and a half years since I came here the last time. I didn't even remember I was registered here. Although not to that level of detail, I already have an idea of how the thing goes. But thanks anyway!

Amazing...