[VOIPSEC] Voipsec Digest, Vol 14, Issue 17

[Tyler Johnson]

We have a concern that the whole SSL VPN approach may be marginally 
acceptable for voice (small packets) but fall apart with video (large 
packets) due to queuing issues. And so there is a danger in heading down 
this path thinking that we have a solution and then it falls apart 
quickly as folks do more multimedia.

That's why we think end to end authentication and encryption by the 
endpoints is the right architectural path from a security perspective. 
That does, however, leave the NAT problem open, but I question the 
wisdom of using a security solution (IPSEC, SSL VPN, etc.) to solve an 
addressing problem. Sure, it might by accident, but not by design. That 
seems like a setup for problems to me.