[PhoneBoy]

Please comment on this article here:

[jbond]

How tiresome the SIP community are becoming. It's a marketplace out there. If you can't compete with Skype, moaning about "non-standard" and "Proprietary" won't change a thing. So come up with something better and go and compete. I don't want to see Skype or Microsoft or Yahoo dominate consumer VoIP-IM with a proprietary solution. But I do want a system that works and I want it now. And currently the SIP community can't provide anything that works as well as Skype.

[DaveP]

Sounds like your mind is made up -- Skype wins! So why are you here instead of in the Skype forums?

[jbond]

Because I read the original article via RSS via Topix.net and wanted to leave a comment. And the only way to do that was to join the forum and have the comment disappear into here.

I don't know if Skype wins. But it's winning now. So what is the SIP community doing about it? Except moaning?

[mberlant]

What, exactly, is Skype winning? Is it the competition to be the best instant messaging system?

This article does report nothing but hype, since all of the tools mentioned in the article are either already implemented or readily available to all SIP users. Bowing to this hype, it tries to compare the apples of an instant messaging system using a proprietary protocol with the oranges of SIP, an open telecommunications protocol.

It is funny that Skype has managed to position themselves as a competitor to SIP, apparently convincing many people that they are a telecommunications service provider, rather than the instant messaging system that they are. That is truly amazing.

[jbond]

If you look at Skype as a PC to PC to POTS to PC VOIP system, there are SIP competitors. And the quotes in the article are from XTEN which is probably the market leader in that area. And frankly XTEN sucks in comparison to Skype as do the others. This is a big market and it's the one that my comments are mainly aimed at. Skype has raised the bar and the pure SIP competitors need to put up or shut up. Maybe Gizmo will do it, but they're now up against an incumbent. They will have to execute at least as well as Skype and market better. I'm not sure they can do it.

If you look at Skype as IM with VOIP capability and compare it with other IM systems with voice. It's better and it's rapidly catching them in market share. There are numerous stories of people who use IM, install Skype and find themselves switching and turning off their old IM systems. It's at least as good but has a few features that blow them away. And the voice "just works" where the other IM voice functions typically "just don't work".

If you compare Skype with telephone replacement via VOIP such as the area Asterix or Vonage is in, it clearly has some problems. Not least of which is the need for a PC in the process and the lack of interoperability. It is possible to compare them and Vonage in particular doesn't look that good in comparison. And even here, access from a PC and IM style presence and chat functions are desirable. And they're usually provided by softphones such as XTEN. At which point we're back to the first market I've described above, and the problem that the SIP applications suck.

I think the big point I'm trying to make is that the proprietary vs open protocol argument is irrelevant if the open protocol implementation is no good and can't get market and mind share. Perhaps the SIP community should be revelling in Skype spreading the message while plotting to blow them away with superior function. But all I see is moaning.

[mberlant]

I agree with most of what you say when you compare a softphone SIP client with Skype's proprietary client. Despite the fact that I have found it impossible to get Skype's client to work with my Linksys router (it crashes my PPPoE connection, requiring a hard reset), I recognize that most of their users have no such problems. Your tale of people migrating from other IM systems to Skype also resonates with me - I switched away from AIM and MSN three years ago when Yahoo Messenger became the first to get their voice and video add-ins working well. To this day, Yahoo Messenger's voice and video functions "just work".

I diverge with you when you imply that SIP users are predominantly softphone users. I believe that more SIP users around the world use hardware ATAs, which allow connection of an ordinary telephone, than use softphones. I base this comment on the observation that most commercial SIP service providers provide hardware adapters to their clients, and many only support hardware solutions.

As far as ease of implementation, I could go as far as to call that a tossup between the installation of Skype on a typical PC (by a reasonably competent user) and the installation of an ATA received from a SIP service provider (by the same reasonably competent user). While we hear mainly from people with difficulties here and on the other various support forums, I don't believe any system, IM or SIP, could survive without the vast majority of their respective subscribers being trouble free.

I also don't see supporting evidence that "SIP applications suck". I have not experienced any troubles getting either XTEN or SJLabs products to work on any Windows machine I have tried. Contrary to my experience with Skype's client, X-Lite and SJPhone just work.

Your last paragraph talks about proprietary vs. open protocol. You are right in that form follows function. Instant messaging systems have tended to use proprietary protocols because the nature of market differentiation in the user experience has dictated that course. VoIP providers have coalesced around SIP, forsaking both H.323 and MGCP in the quest for interoperability and reliability, believing that a uniform (PSTN-like) user experience is most desirable for their market.

I also don't happen to believe that Skype has superior function to SIP. I value the seamless interoperability with the PSTN and the way a telephone connected to an ATA behaves just like it would if it were connected to the PSTN. And, apparently, there are many people (especially in Texas) who are so at ease with the telephone-connected-to-ATA solution that they have coerced the FCC into mandating E911 support for users in the US (even those of us who neither want the feature nor want to pay for it). That's pretty strong evidence in my book.

[PhoneBoy]

The things that Skype has over SIP right now are:

1. NAT Traversal. If there's any hole in your firewall, Skype will find a way through it.
2. Integrated IM.
3. Multi-platform support (Win/Mac/Linux/Pocket PC)
4. A nice UI that is consistent across platforms.

None of these things are insurmountable, though. The P2P SIP stuff sounds like it could be way cool once it is implemented in a usable cross-platform product. It remains to be seen what that product will look like and when it will come out.

[sethg]

This is an old thread, but there are issues here that are still relevant to the way that different VoIP solutions work. As a hardware engineer, primary in my mind is the issue of latency and it's evil twin, jitter.

An outbound-only provider who terminates SIP/IAX calls to the PSTN will typically have their own private data network that covers most of the distance from VoIP caller to PSTN recipient. While on this private network, latency and jitter are relatively easy to control. At some available gateway near the PSTN recipient, the gateway places a PSTN call to the recipient. That also has very low latency, as the call is carried by ILEC's or CLEC's. The only segment that is uncontrolled, for the great majority of users who do not have direct NOC connectivity, is the originating leg of the call where it has to traverse the public internet. Even though this may only be a few hundred miles, this is where the bulk of the latency/jitter often occurs. The reason is that this traffic is routed as IP data traffic, even when encapsulated over ATM for short-hauls by your local telco. IP data traffic has no QoS guarantee, different packets in the same session do not have to travel on the same route, the packets are allowed to arrive out of order and return packets often do not traverse the same route. With all those caveats, these things can happen but generally don't. This explains why it works at all, considering all the things that can possibly go wrong in that short voyage through the public internet.

What does this have to do with Skype or P2P SIP? Quite a lot. Skype does not have a private data network, that I am aware of, and everything is routed over the public internet. While it generally works very well (I have been both a Skype and SIP user for a while), sometimes it works horribly, and I attribute that to the particular routes chosen by internet backbone routers. If you hang up and call again, you will often have a completely different experience. In short, Skype calls are routed over the public internet end-to-end and are subject to all the problems inherent in that. I'm sure the folks at Skype had to work a lot harder to deal with latencies that abruptly change during a call, far end echo characteristics that change as the routes change and jitter that is not stable during the call. It's quite an achievement on their part. Too bad the code is closed source so that the SIP community could benefit from their advances.

As we consider a P2P version of SIP, realize the the call will travel over the public internet just like Skpe. The outbound VoIP to PSTN providers that carry your call on their private data networks are not going to do that for free just because you are terminating on a SIP/IAX endpoint. They could certainly afford to reduce their rates for such calls, since they don't have to pay for the final hop over the PSTN. But my understanding is that P2P SIP is supposed to be direct connection from one SIP endpoint to another routed over the public internet. That being the case, the SIP community will have to do everything that Skype has done to make such calls usable. That would take a lot of effort as well as a lot of focus. In that regard, a single, well-funded enterprise like Skype has an advantage as they can decide on an approach and devote enormous resources to it. A large open-source project like Asterisk has a shot at accomplishing the same goals, but a disjoint collection of small entities are not likely to do it.

As an example, consider the NAT traversal problem. This is hardly rocket science, but Skype has come up with a robust method of penetrating NAT routers with no user intervention in the vast majority of cases. In comparison, the SIP solution is a collection of things that work in particular cases. Sometimes you need to set up a STUN server, sometimes not. Some routers will need port forwarding, and not just TCP 5060 but a collection of UDP ports that is application specific. This is pitiful, it is a black eye for SIP and frankly, there is not much excuse for it.

I greatly favor open vs. closed protocols and thus I am very much in favor of SIP predominating over Skype, MSN, Yahoo and anything else closed. However, if the SIP folks are not willing to take on the large problems that Skype had to overcome to deal with calls routed completely over the uncontrolled environment of the public internet, we probably shouldn't be seriously talking about P2P SIP.

[edited for spelling/punctuation]

[PhoneBoy]

Please comment on the following article: Is P2P SIP Poised to Out-Hype Skype?