I consider VoIP Supply in the US one of the best VoIP on line shops on that side of the Atlantic. I really wish they had a European presence and when they decide carry a product I trust their judgment.

They also have a blog called VoIP Insider that is worth following. They have manged to walk the fine line between self promotion and really newsworthy information within the VoIP cloud.

When these guys say that they are going to sell a certain device - it's worth taking a closer look.

So when I learned that they are going to sell a IAX based hand set I became very interested for several reasons which I will address later in this posting.

A contest worth participarting in

They are so confident that the Citel C4110 is a excellent buy, they are willing to give away 3 phones (and t-shirts). They are currently running a contents which is not really hard to get into: Just go over to their blog entry called Win one of three new IAX phones! and write in the comment field the difference between IAX and SIP.

I must admit that this is the one of the few contest I have bothered to take part in for the last 20 years. The reason is the ability to get my hands on a IAX handset of good quality.

Not only does the spec looks good - the phone it self looks sexy. I want one!

IAX telephones are fascinating creatures

Over the years I have tried a lot of handsets and ATAs and ITAs from the Far East - and most of these devices are prone to the Far East Syndrome. The short run down on this syndrome is that even if a unit is extremely interesting at first sight, it may have a great deal of shortcomings making the device very unusable.

I fondly remember a DTA (Dect Terminal Adapter) which not only had DECT built in. Throw in WIFI, a WAN port and 6 LAN ports + simple firewall capabilities and you had the best product that year. It even had a analogue port which could be set to FXS or FXO depending on if you wanted to have a backup line over the analogue network or plug in a FAX or a analogue handset. DECT is big in Norway - so I was more than excited. Given how much DECT is deployed in Norway over the last 20 years this product could be a killer. Everything worked as expected - except the DECT part. The DECT part of the box could NOT route DIDs to given hand sets. We could not seriously go into the market with such a device when DECT users where more than accustomed to route DIDs into specific hand sets.

I have tried several of the IAX based hand sets you can find on eBay. They are cheap. They look good. Some even smell good when the package is open. But in all the hand sets I have tried it boils down to the Far East Syndrome. One or two details that are so severely hosed making the device just painful to use.

Even if I have not tried the Citel 4110, but only seen the spec, it is carried by Voip Supply. This it self should be a indication of good quality.

Why IAX? Is not SIP the solution to everything VoIP?

There are three main reason why I think IAX is a better protocol to deploy in certain environments compared to SIP + RTP.

The first reason is the firewall issue. For users with simple firewall setups (”nothing goes in, everything goes out”) SIP can, in theory be as easy to get going as IAX. However in practice this is often not the case. The simplest part of SIP is the protocol running on port 5060. This is seldom a issue, but as soon as you involve RTP, problems will sooner or later rise and create a lot of confusion and iritation.

When I was working at a (now defunct) ITSP we had some big customers with some “real” firewalls. More than once security personnel “tuned” the firewall and put a too strict limit on the number of ports open for RTP. Who got the blame when the VOIP service did not work? Yes - the ITSP.

In comparison the IAX protocol only need one port. Everything is tunneled inside the the IAX protocol: call control information and the voice data itself.

This brings me to the my second reason on why IAX may be better to deploy in certain environment: Overhead.

Yes - network overhead is an issue.

It has been a while since I checked the numbers, but as far as I remember the overhead using RTP is around 90 bytes per voice packet compared to IAX. May not seems like a great deal - but do not remember that in one second speech there can be as many as 50 packets (20 ms each). So we now have 4500 bytes overhead pr second! Maybe not very much for a small company - but for a big company with hundreds of concurrent calls this adds up. Seen from the technical point of a ITSP - usin IAX make much more sense than to use SIP.

Of course, from a business point of view SIP is the major player.

The third reason is not necessarily very valid - but it is worth taking into account: IAX is so much easier seen from a implementation point of view than SIP + RTP.

Granted, SIP can be used, and is used, for much more than IAX is today (video springs to mind).

But SIP has become extremely complex. Just count the number of RFCs.

One of the reason SIP was invented or created was because H.323 was complex.

I have not much experience of H.323 except to not like it. If SIP was a supposed to be a simplification compared to H.323 - the the current crop of SIP RFCs are not proving this. I have read more than one place that SIP is now becoming as complex as H.323. What an ironi.

SIP is a very good example when you let people without knowledge of Internet (traditional telecom people in this case) try to create a Internet standard.

I know I only had 3 reasons - but for added bonus I’ll put in a 4th reason.

Security.

Security on SIP is not straight forward compared to IAX. I have observed more than once that VoIP (that is SIP based) was not possible to deploy even inside a organization due to internal firewalls and internal security policies. I am not talking about SRTP or SSIP as being the problem solver - just the number of ports that need to be open.

Even if one enable encryption on the IAX protocol, it’s still running inside it’s own tunnel. In fact - enabling encryption on IAX does not implicate the need for amending firewalls the same way that one need to do if using SSIP and SRTP.

A story from CeBIT in Hannover this year to prove this very point: My company have created a secure call solution which uses SIP + RTP. For all practical purposes we emulate the tunneling features of IAX with the added encryption thrown into the equation. One of our visitors to our CeBIT booth got a presentation of our solution and exclaimed that: “Wow - no more firewall problems like we now have with our customers”. It’s a sorry state for a protocol suite when one have to tunnel everything to get around network roadblocks for voice communication.