Tuesday, April 04, 2006

Skype, Asterisk (and IMS?) (and P2P SIP) cont. 

So we have Skype, we have Asterisk, we have IMS, and we have PoSIP (plain old client-server SIP) and we have the nascent P2P SIP. What will be the future?

First some lessons from the past:
  • Walled gardens are very successful for some time, but do not last.
  • Proprietary solutions are also successful for some time, but they are either replaced by an open standard, or they evolve to a de-facto standard
  • Open systems (open source) are lasting and getting more and more sucessful.
  • IETF standards are more sucessful then others
  • End-to-end systems (the Internet) with distributed intelligence are replacing centralized Intelligent Networks
  • It is essential that third parties may add functions
  • Simplicity wins, also usability.
  • The ultimate end-to-end system is P2P
I know one cannot compare these different applications and protocols consistently, but let's give it a trt:

Skype is a walled garden, it is proprietory, it is definitely not open source, but it is P2P, it is end-to-end and it has an open API. It is simple to use, eventual complexity is hidden.

Asterisk is open source, it is using a propriatory protocol (IAX2 - input to IETF, but in competition with SIP), it is client-server, not P2P, it is getting more and more complex. Asterisk is a provisioning night-mare.

Client/server SIP (PoSIP) is open standard, it is partially end-to-end, but is getting more and more centralizez and drawn into walled gardens (SBC, IMS). It started simple, but it is getting also more and more complex. It is not easy to use (configure).

IMS is the a showcase of complexity. It is not necessarily in a walled garden, but all use-cases go in that direction. It is an open standard, but the question remains how third parties may implement additional functions. It is in essence a direct replacement (or migration) of circuit switching networks including Intelligent network (IN) functions with IP technology. It is not end-to-end and it is not P2P.

P2P SIP is an (or will be) an open standard, it is definitely designed to be end-to-end, it has by definition none or only rudimentary central intelligence. One design goal is self-configuration, i.e. no provisioning. Is it the last stand of IETF against Skype and IMS, or is it leading somewhere?

Predictions are difficult, especially for the future (even if it is already here, as some say ;-)

I have a dream:

Let's assume, I take Skype with all its benefits (P2P, end-to-end, nailing through most of NATs and firewalls, end-to-end encryption, usability of the client, ease of installation and provisioing, ...) and remove all problematic stuff: proprietary protocol, walled garden, no interworking with the rest of the world, except via a very limited API.

One serious drawback of Skype is that you always need a PC running. I would e.g. like to have a Skype client in a box. As I said, I tried yesterday the EQO add-on to Skype. This would be a real killer, if I would not need to have my PC running. Since my PC is basically my little laptop which is always with me, this nice application is somewhat useless, because I either have my laptop on and could use it for Skype calls anyway, or it is off, and I cannot use EQO. What EQO does is basically setting up a conference call where one leg is on the PC. So EQO it is nice, but useless (for me).

I understand that one reason is the processing power needed for Skype, the other is the closed application which does not allow an open source implemention on Linux devices.

Now if we could replace Skype with P2P SIP applications with all the benefits, but without the drawbacks, and in addition using client/server SIP to interwork between them, we could implement many little interoperable "Skype" islands. All the add-ons for Skype could still be used.

All kinds of real-time communication could be implemented on top of the basic Internet transport end-to-end. Unique identities are still provided on one side from the P2P networks (the user-part of the SIP URI), and the domain-part is provided via the DNS as is.

Add on top of this certified identities (this still has to be somewhat centralized) and keep them completely separated from the underlying communication, the SPAM and SPIT problems would also be solved.

So I see a future in P2P SIP. And I am also sure that there is still a place for Asterisk in this scenario. But I do not see a place for IMS.

I don't see any future for SIP. Short term - yes. Long term - no. There is more to messaging than only VOIP - and currently VOIP is the only wide-spread SIP application. I believe the future is in XMPP/Jingle (the same protocol which is also used by Google Talk and which is currently being standardized by the Jabber Software Foundation). It is in some way similiar to SIP, ICE must be implemented and the combination of XMPP/Jingle supports more or less the same features which Skype already supports (which the exception of still requiring a server). SIP was a nice idea - but all the extensions and hacks make it rather ugly. Jingle (IMO) looks perfect so far.
Nah, if you don't see a future for SIP means that you are not working on this industry.

Did you know you could use SIP for messaging, IPTV,....? SIP has nothing to do with voice, it can be used for anything, one could think of using SIP to browse the web if it was just one layer below ;).

Which are the ugly extensions and hacks? do you really think TCP/IP spects were written in one go?

I think you both are forgetting the main actor from the movie the $ or businesses/companies involved, at the end of the day is them who will have to get something in return or do you think Google is going to pay their bandwith bills?
Any one want to hazzard a guess at the top SIP applications? I see all kinds of features and combinations but would like to narrow down a list of potential killer apps (SIP) based ... and preferably Non-IMS.
Maybe a HAZARDOUS spell checker! LOL
Nice blog you've got here. Keep up the good work. I am a blogger too, specialising also on VoIP technology.

My blog receive good traffic and the index and links page have Page Rank of 5. You can check out my

blog at www.solokay.blogspot.com .

I am wondering if you will be interested in a Link exchange or Blogroll entry exchange from our index or links page. Being high rank websites, linking together will considerably increase our link popularity and also drive more traffic to our blogs.
Moreover our readers will be exposed to more quality contents they might be interested in.

If you are interested you can use this details to link to me:

Anchor text: VoIP Telephony
URL: www.solokay.blogspot.com

You should send your links details to me too so that i can add your link immediately.

Hope to hear from you soon.

Yours sincerely
Solomon Ige.
As far as I see the XMPP/Jingle train it is another crack of the "walled garden". This is a conflict between innovation and different industries - telco vs. non-telco like Google who originally developed Jingle. By the time NGN adaptation of the telco industry completed the edge centric Internet industry will be so much further ahead.
I was just wondering if you'd mind telling me a bit more about whats happening with p2p sip?From what I can gather it's currently in the process of being finalised into a formal standard based on the existing internet drafts - is this the case?has any timeline been given on this?

Many thanks in advance for the info.
Post a Comment

This page is powered by Blogger. Isn't yours?