What you are describing is similar to the Acme TSM product where sip / rtp are wrapped in a tunnel for nat traversal purposes. It actually has quite a few strengths but all of them depend on the tunneling devices ability to detect failure and re-negotiate. If there are any interfering devices in the middle with ALG's etc it should cut right through them, and as far as overhead goes, as long as you arent doing any crypto its actually pretty light. You will want to be aware of path MTU for signaling traffic since the extra bits of overhead can pad a large sip packet (think invite with custom headers and large-ish SDP) over 1500 bytes which will cause fragmentation issues. Consider going to compact headers to eliminate this and make some overhead room. RTP wont have this issue of course. Will you be doing sip over udp or tcp for this? Who will be managing the GRE tunnel endpoints, and what devices will be doing the tunneling? What is your sip redundancy model right now? On 08/13/2013 07:17 AM, Bob Hancock wrote:
I have a project in which a user is interesting in interconnecting and deploying ATAs for VoIP services but wants to use private IP addressing on the devices. My initial though is to NAT these devices and allow my SBC to handle NAT traversal for them. This allows the to use their dedicated connection to my network or if that link goes down fail over to moving to connecting across the internet. They are interested in potentially doing a GRE tunnel and keeping the private addressing in place as it traverses that tunnel. I am concerned about encapsulating the VoIP traffic as well as with the fail over options. Does anybody have any experiences with VoIP over GRE tunnels and what type of results and challenges you encountered? Thanks.
--BH
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops