Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

Let's imagine a company that doesn't yet have a "professional" video conferencing solution so they use skype.

Now, let's imagine that this company has lots of branch offices, many of which have one or no public IP addresses.

Are there going to be problems using skype as the video conferencing solution between a branch office with no public IPs and another branch office somewhere else in the same RFC1918 space?

I can see one situation where the skype client is clever enough to go over the private backbone, but I can imagine other ways in which it would work very poorly because it will be doing all sorts of nat traversal and other stuff that depends on scarce resources like the common external gateway over a potentially laggy line instead of the obvious direct path.

Does skype simply work best when you've got two nodes with dedicated and real public V4 addresses?

Anyone have any direct experience with this?

share|improve this question
1  
It sure seems likely everything will be fine. Skype is primarily used by consumers and most of them are behind some really cheap broadband NAT-router. –  Zoredache Mar 1 '11 at 17:20
1  
@Zoredache: It sounds like the OP has branch offices that are connected to the main office by WAN lines but share that main office's Internet connection. I think it's likely that calls between clients will probably end up routing out over the Internet and back even though they're part of the same WAN. I can't imagine Skype was designed to work well in this kind of scenario, but the OP will probably have to monitor it to see. –  Evan Anderson Mar 1 '11 at 17:34
    
@evan right -- I'm worried about hairpin traffic going over a laggy (for the BO) link instead of directly from 10.0.0.0/8 to 10.0.0.0/8. –  chris Mar 1 '11 at 19:43
    
My gut says that you're not going to get the results you want, but try it and see. Sniff the traffic (or just use "netstat") during a call and see how the endpoints are "talking". My guess is that the clients are registering with a "supernode" outside your network using their public IP address, and that conversations will end up running thru that external supernode. (This is not unlike using "LogMeIn" from a PC to a PC behind the same firewall, just with the Gnutella network instead of LogMeIn's persistent TCP connection mechanism.) –  Evan Anderson Mar 1 '11 at 20:37
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.