Heading out to Los Angeles this week for Communications Developer Conf / ITEXPO...
Google's Android and the future of the (open?) mobile Internet

Skype and SIP interop - the two sides of the issue raised by Michael Robertson

skype_logo.pngShould Skype open up it's network to other users? to other networks? Should Skype stop preaching about "openness" when it's network remains closed?

In the middle of last week, there was quite a little storm raised in the VoIP corner of the blogosphere after Andy Abramson published a letter from Gizmo Project founder Michael Robertson critical of Skype's openness after Skype continued to call upon the FCC to open the wireless network to applications. (See also here and this Skype blog post (and this one) for background.) Being at ITEXPO last week, I didn't have the chance to blog about this at the length I felt it deserved until today.

First, for some context, here are some of the blog posts last week:

All of it makes for good reading. It's an important issue.

So I guess I'm somewhat of two minds on this issue.... while I agree with some of Michael Robertson's points, I guess I see the whole issue of "Skype openness" as quite orthogonal to the larger issue of open wireless spectrum. I'll write about both issues at some length below.

This is long. Don't plan on reading it on a crackberry or iPhone. Best get a cup of coffee and read it on a laptop or something like that.


To put it another way, I completely applaud Skype's letter to the FCC and think that the battle for opening up the wireless networks is probably the preeminent "battle" we who are advocates for an open Internet have before us. The wired carriers are well on their way to being commoditized big, fat, dumb pipes. The telcos started selling "data lines" and then soon the world of IP wound up increasingly removing them from the picture as anything other than a big pipe... and that's not the world they want. They are fighting (and will continue to fight) to retain relevance (and ARPU - Average Revenue Per User) but the IP train left the station a long time ago and the telcos are scrambling to catch up and stay on board rather than just being (some of) the providers of the track.

On the wireless side, not only do the carriers own the track, but they still own the train and they are still driving that train. They control who gets on and off... how fast or slow it goes... what color the train is, etc.

I, for one, don't want that. I want them to give us a solid set of tracks to use... but I want the train to be open to all. I want the wireless carriers to be big, fat, dumb pipes. I want choice. *I* want to be in control.

The carriers naturally don't want to relinquish this control. They see how they missed it on the wired side. They want to keep the wireless walled gardens for as long as they can.

The cracks are appearing... Apple's control over the iPhone and AppStore is a phenomenal crack in the telco walls...... although it ultimately really means exchanging the walled gardens of the telcos for the bright shiny walled garden based in Cupertino, CA. I'm not sure that ultimately is the best for all of us... but it does crack the telco walls. I think Google's Android has more potential... but we'll have to see later this month when those phones first come out.

So with that view, you can expect I applaud Skype's efforts to open up the wireless networks and allow consumers to have a choice of what apps they want to run. I want the *wireless* carriers to be big, fat, dump pipes... give me an IP address on the *mobile* Internet and let me do what I want with it. Sure, the carriers can offer their own services, and maybe if I like them I'll pay for them.... but I want the option to use other products and services - without degradation or prioritization...

To put it another way, I pay the wireless telcos for *dialtone* now. Once connected, I can call anyone and use any *voice service* over the PSTN. I could use someone else's voicemail if I want (like GrandCentral), although the carrier's offering may be more convenient (and is usually free). But I can call anyone on the PSTN and use any voice service I want. The carriers just provide me dialtone.

I want "IP dialtone". I want a Big, Fat, Dumb Pipe.

So... go, Skype, go!


Yet having said all this, I agree with Michael Robertson that Skype's got its own issues with openness.

I don't like walled gardens. Period. End-of-story.

I don't like telco walled gardens. I don't like Apple's walled garden. I don't like Facebook's walled garden. I didn't like the walled gardens of CompuServe, AOL, Prodigy, Genie, etc. and I rejoiced over time when the open standards of IP tore down those walls and brought about the Internet (with all of its warts) that we have today. [Tangent: I do worry, actually, that we are retreating a bit back into the walled garden world with things like Facebook and Myspace... but that's another topic I've blogged about.] I can somewhat see some value in walled gardens during the early stages of a product or technology as it reduces the set of parameters/variables and allows the service to be stablized/fixed/improved. But at some point the walls need to be reduced/eliminated. As a security guy, I don't like monocultures... I don't like homogenous systems (where one virus or issue could wipe out the whole system)... I like diversity... heterogenous systems. I don't like single-points-of-failure. I don't like single companies (or small sets) in control. I don't like walled gardens.

I don't like Skype's walled garden.

The PSTN run by the telcos of today does not provide the rich communication experience we want. We need to bypass it and leave it behind and build the massive interconnect of IP communications systems. Players like Skype have a key role in my opinion in building that new infrastructure.

But if we exchange the current PSTN walled garden controlled by the telcos for a new walled garden controlled by eBay/Skype, have we really gained anything?

Sure, it gives us a rich, multi-modal user experience. Sure, it's nice and pretty. Sure, it gives us a central user directory. Sure, it gives us wideband and encrypted audio. Sure, it's cool and all... but it's still a walled garden controlled by a single company.

Ultimately, I would like to see a new voice infrastructure that consists of many different "clouds" all interconnected and able to communicate between the clouds. Skype is one cloud. So are the SIP clouds being run now by various carriers. So are the Voice-over-IM clouds like MSN, AIM, etc. (that try vainly to compete with Skype). So are the various systems being built by vendors all over the place (including all the Microsoft OCS clouds starting to appear within enterprises).

We need to build the interconnect.

Yeah, there are a TON of issues out there that we still need to address to build that interconnect. There's a whole host of security issues... there are billing issues... there are trust issues... there are network plumbing issues. Yes, there are all those issues. But if we are to succeed in ultimately bringing about the rich communication experience we want, we need to make this happen.

And for that, Skype's walls need to come down.... at least a bit.

What we need is that Interconnect from Skype's cloud out to the emerging IP infrastructure. Think about it... Skype right now has a two-way interconnect between Skype's cloud and the cloud we know as the PSTN. It's called "SkypeOut" and "SkypeIn" (or whatever marketing names they are being called now). If you dial my SkypeIn number, you can reach me on Skype wherever I am. From my Skype client, I can call anyone on the PSTN. The two-way interconnect is already there.

So why not offer the same on the IP side?

Because I work at Voxeo and we were one of Skype's original Voice Services partners, I already know that Skype has a massive SIP infrastructure on the backend to do the the PSTN interconnect. Skype users can even dial a specific +99 number and the call goes from Skype's SIP cloud over to Voxeo's SIP cloud... and it works beautifully.

So one half of the interconnect is already there - although only for limited numbers of partners.

But where Skype already has the infrastructure, why not look at making that capability more accessible? What if someone in Skype could just type "sip:[email protected]" and the call would go out from Skype's cloud to the service providers? This could be a new feature as part of the unlimited calling plans, etc.

How many people would use it right now? Probably only a tiny few... today. But suddenly Skype becomes an enabler of the broader post-PSTN infrastructure. New companies can get their services up and running knowing that they can promote them to Skype users and have Skype users get to those services.

Plus, Skype can now connect to all those enterprise VoIP systems being deployed everywhere... so for all those IT managers blocking Skype now but allowing SIP gateways for remote teleworkers using IP phones... Skype can suddenly be that remote softphone being used in the sense that it could connect in to other people on the corporate enterprise system - they just become "sip:" entries in my Skype directory. Skype still is my overall directory and user agent.

And what about the other way? Wouldn't it be great if someone out there on a SIP system could just call something like "sip:[email protected]"? The call goes from their SIP cloud across the Internet to Skype's SIP gateways and into the Skype cloud.

The SIP system user can do this right now... via my SkypeIn number... but they have to use the crappy PSTN. Why not ditch the PSTN and go directly across the IP infrastructure? Hey, maybe some user of a "HD Voice" Polycom phone could call Skype's gateways via SIP and actually wind up talking via wideband audio? (Yeah, okay, I'm probably dreaming on that since Polycom supports G.722 for wideband and Skype uses its SVOPC codec.)

I personally would probably wind up using my Skype client more for a simple reason that I have a SIP IP phone on my desk... but I'm not at my desk all that much. Wouldn't it be great if I could forward that to the SIP URI of my Skype client? (Which I can do now by forwarding to my SkypeIn # but again I'm going across the crappy PSTN.) Or better yet because I have a SIP URI for Skype it becomes one of the various phones I ring when someone calls that number (it's not, now). The number on my business card would then wind up actually going to my Skype client.

Suddenly Skype can be a player in the enterprise "unified communications" market. People don't need Skype-to-PBX gateways or sacrifice chickens and utter weird incantations to get Skype connections working with open source VoIP systems. Skype users can talk to Microsoft OCS systems... or Cisco IP PBXs... or Avaya's or Nortel's or Mitel's or ShoreTel's or... or... or...

What if... even... I could do a SIP invite to make a video call to another system? (Okay, so now maybe I'm really dreaming...)

Suddenly people (like Michael Robertson) have fewer reasons to complain. Sure, the Skype client still uses it's own proprietary protocols and codecs for communication within the Skype cloud... but you can interconnect.

Suddenly Skype is a leader in building the broader overall next-generation IP communications system. Skype's not a walled garden but rather a player in the larger picture.

Sure, there's a whack of issues involved with doing this. On the technical side, Skype has got to build SIP gateways that could deal with the abuse they would undoubtedly suffer by being exposed on the public internet (like any or all of the VoIP security tools out there). They have got to make sure such gateways don't become a way to inject spam/SPIT into the Skype network. Skype has got to figure out how to package it... potentially charge for it, etc. And they have to deal with all the glorious interoperability issues that come with SIP... as the protocol increasingly becomes an unmanageable accretion of all sorts of crap. (And I say that as an advocate for the SIP protocol.)

Ultimately, I think that's the kind of openness Skype needs.

Skype needs to provide the same two-way interconnect to the evolving IP communications infrastructure (that is almost all becoming SIP-based, for better or worse) just as Skype provides the two-way interconnect to the PSTN.

Build it and Skype would silence many of the critics of Skype's lack of openness[1] and give Skype a (much-needed, IMHO) hype-boost among the early adopter crowd who also plays with all the other emerging tools. It would be a bold move that would also help Skype gain some credibility and recognition within the larger industry. In my opinion, outside of the technical issues it would go far in so many ways in helping Skype grow - and helping the industry grow.

Will Skype get there? Good question...

[1] Not all critics would be silenced naturally because someone would still complain that they can't connect their particular client to the Skype P2P overlay network. Or that they can't connect XYZ hard phone to the Skype cloud, etc., etc. But it would silence many of the critics of Skype as a "walled garden".

Technorati Tags: , , , , , , , , , , ,