Thursday, January 26, 2006

You Only Live Twice - The Death and Life of UDDI

The death of the UDDI Business Registry was far from being natural. This post restores the dramatic moments of an untold story.

Disclaimer: in this post the characters and events are either the product of the author's imagination or they are used entirely fictitiously.

With Web2.0, the vision of the Internet-as-a-platform has become omnipresent, and its vocation understood. The ultimate goal of the Internet-as-a-platform is to create a universal, friction-free, flow of Information. As Thomas Baekdal correctly said, "The 'internet as a platform' is going to be replaced by the 'internet as a pipeline'".
In a pipelined world, the success of software companies – i.e. the pipes - is measured not by the amount of product downloads (aka, end-user clients), but rather by the number of times their Information has been reused, distributed and aggregated. Under this paradigm, there's no longer a distinction between a server and a client, a provider and a consumer: humans, machines and devices are all pipes in a platformed world. And the bottom-line is that APIs - the enzymes carrying the Information - will be prioritized over end-user clients (Ajaxed or not), as the later are stopping the Information circulation, while APIs enable reuse, syndication and distribution. Briefly, APIs are the bread and butter of a company's eco-system, and the following self-explained pictures and article by Phil Wolff from excellently demonstrate it.

This leads us, inevitably, to an essential, global, infrastructure service: the Directory.
As Alec Saunders concisely defined in his voice 2.0 manifesto, the three pillars of the future are "presence, directory, XML". Internet Directories, maintaining a list of mash-able APIs are a meeting/starting-point for entrepreneurs, innovators and simply bored geeks. If you want to participate in the Information creation and circulation, the APIs Directory is a reasonable place to start with. The current Internet Directories range from sheer yellow-pages, like programmableweb, through vendor-oriented Directories like the recently launched's AppExchange, to universal marketplaces such as StrikeIron.

I was therefore highly surprised to learn that IBM, SAP and Microsoft (the first two being "off-web" for a dangerously long period of time, and the later trying hard to get on the web2.0 platform) have decided last month to turn off the lights on their joint UDDI Business Registry (UBR) – the "father", as we like to say, of Internet Services Directories. The Trio claimed for a natural death, stating that:

"The primary goal of the UBR was to prove the interoperability and robustness of the UDDI specifications through a public implementation. This goal was met and far exceeded… …The UDDI Business Registry (the public UDDI demonstration registry) being discontinued."

My first reaction was – Guys, haven’t you heard about refurbishment? I mean, why throwing away an old veteran which can well be used as a foot in the web2.0 door?

Intrigued by this obscure logic, I decided to sniff around. First, I rechecked the UDDI goals, running through some historical documents. Here's an excerpt from October 4th, 2001:

SAP Fully Embraces UDDI. SAP Becomes UDDI Business Registry Operator

SAP will fully leverage UDDI for e-business solutions-- from providing SAP functionality for Web services to using services published through UDDI… SAP will make it possible to provide the application functions of the e-business platform as a service, which makes it easy for customers to provide this information externally through UDDI…

So UDDI has not been conceived as a demonstration registry, but rather as an active marketplace. I then checked how successful this marketplace has been:

Chris: "I'm actually a bit surprised that this announcement (the UBR Shutdown) didn't make more headlines. The real news here is that a UDDI registry still exists!"

Steve: "Like UDDI ever made any sense…If you are a service consumer, with your line-of-business systems dependent on remote, outsourced services, do you want a UDDI registry finding endpoints for you, or do you want a long lasting relationship with somebody you trust?"

Clearly not a hit...

I then told myself, that probably this ShutdownFAQ published by IBM is IBM's own interpretation, and that checking out what Microsoft and SAP say about the shutdown might yield other perspectives. But – surprisingly – both Microsoft and SAP Shutdown FAQ pages returned with a HTTP 404 kind of message. Is someone here destroying evidences?

UDDI, as we all knew, has been clinically dead for years now. So why did IBM, SAP and Microsoft decide to disconnect it now from the life-support machines? Unless this was not an Euthanasia… according to the bill of evidences collected so far, I suspect it was a cold-blood, first-degree, murder.

The UBR crime scene looks like this: a body; three suspects; lack of alibi (why disconnecting now, against all common web2.0 sense); two lies ("met its goals" and "demonstration registry"); and evidences removed from the crime scene (Shutdown FAQs removed). The only thing we lack now is - a motive.

My next post, which hopefully will be published early next week, will elaborate on the alleged, speculated, motive. I would hint that my thesis refers to an undergoing low-intensity war between two approaches for SOA Management. And as SOA is the present and definitely the future, there's a great deal of money at stake here. By killing UDDI, IBM, Microsoft and SAP wanted to eradicate an emerging, threatening rival, whose name is no other than… UDDI.

To be continued…

This is the first post in the UDDI series: You Only Live Twice - The Death and Life of UDDI, The Death and Life of UDDI, Part II, The Toolsmiths, The Manager, His Repository and Its UDDI Lover


By Blogger Robert McIlree, at 1:16 PM  

There's a difference between the fact that the registry existed and what ir contained. If it contains things that aren't wanted, it matters little that it exists.

By Blogger Muli Koppel, at 2:28 PM  

Hi Robert

Thanks for your comment.

Clearly, the UDDI Business Registry had no value for us – its potential business customers. But it has been a symbol, being one of the first three WS standards, along with SOAP and WSDL. Any action involving a symbol, is interpreted symbolically – and in the UBR shutdown case you could find many voices out in the Blogosphere interpreting the shutdown as "yet another nail in the WS-* coffin". As no one - especially not the Gorillas - goes public voluntarily to admit a failure, I believe that IBM et al. officially announced the shutdown in order to gain something – and not to lose. What's this something? I am starting to figure it out now, but that's my next post already…

Robert, thanks for following. I am also enjoying reading your own blog.


By Blogger Robert McIlree, at 4:10 AM  


Thank you for your compliement! I agree that the UBR had symbolic value, but it would have eroded over time if it had continued not to provide any value.

I'll take a totally random guess that the 'something' you're referring to is the selling of granular services - out-services I believe they're called, and I'll leave it to you to post your thoughts...if I'm wrong, then you're post on that will be even more intriguing because I'm drawing a blank past that.

Bob McIlree

Post a Comment

Tuesday, January 17, 2006

A Skype Rashomon: P2P, Voice and the Read/Write Web

Tim O'Brien, a group manager platform strategies at Microsoft, challenged recently the scalability of web2.0 applications: "When you look at Web 2.0 companies out there, they have user bases on order of 300,000. The question is when the base goes to three hundred million, how do you deal with that?"

Undoubtedly, there are several answers to the above challenge. But only one got 2.6B$ (and if all milestones are met – 4.1B$), and that's Skype.

Skype is, for and foremost, a proven (KaZaa, Joltid, Altnet) peer-to-peer architecture that successfully imitates the organic behavior and topology of scale-free (or power law) networks. This type of networks, to which the Internet is a prominent model, get stronger and unbreakable as the number of nodes grows (ok, ok – almost unbreakable; there's always this small Achilles Heel [=the hubs]). And because peer-to-peer networks – just like their step brother (or, rather, step son) of grid computing – are utilizing the endpoints as computing resources, they are extremely cost-effective in comparison with other networks in a need to provide the scalability and reliability that the above mentioned Microsoft's Tim O'Brien is seeking.

Skype architecture and The InternetSimilarly to traditional grids, peer-to-peer networks are usually associated with batch-oriented operations, such as file sharing. Yet Skype is a living proof that mission-critical, real-time services can take advantage of the same underlying peer-to-peer architecture, in order to gain internet-grade scalability and reliability for a fraction of the traditional cost.

Can eBay harness the underlying peer-to-peer architecture of Skype to the advantage of its future Services? My answer is – definitely. I think eBay is building what many expected will come from Google - some sort of a global, infinitely scalable and reliable web2.0 platform. eBay could, naturally, wait for the Gorillas to provide the ultimate grid platform for their services; they decided, though, to go with the proven grid-makers of the Internet. They voted KaZaa.

As said, Skype is another plug-in to the peer-to-peer network infrastructure that the Skype founders successfully implemented in at least three earlier companies. Now, why would eBay want to have a VoIP infrastructure?

One of the official and logical explanations is figured in the eBay-Skype investors' presentation. It appears that 40% of eBay's revenues are attributed to complex transactions that usually require a human touch. This human interaction happens in a form of emails and/or voice calls. In eBay's terminology, this is a friction. Skype will hopefully enable faster Time-To-Deal and remove the friction by allowing natural, intuitive and quick human interaction. As Amazon demonstrated genially with its Mechanical Turk - sometimes it's much simpler and cost-effective to let humans do the job.

I think, though, that eBay needed a VoIP infrastructure for at least two more (unofficial) reasons. The first is related to the future of web2.0; the second – to a viral adoption.
As we know, web2.0 is the social web, in which people are taking active part in the creation of World Wide Information. When technology leaders explain the foundations of web2.0 they usually refer to open source, SaaS – software as a service, scripting, users rating, tagging, rss, ajax, rest and (probably) soap.

But something fundamental is still missing in the people's web, and that's the ability to communicate as people do. The interaction model of web2.0 hasn't changed from web1.0 - it is still browsing and mouse-clicking and not human. Think about any sci-fi movie in which people interact with machines via browsers and mouse-clicks – you won't find any. Voice is the missing protocol of web2.0.

VoIP technologies provide eBay with the ability to create mash-up "humanoids" that would interact with people in the most natural protocol – voice - thus further removing the friction between the man and his mechanical friends (see picture). Actually, eBay (or Skype) has already started encouraging community developers to create this kind of voice-enabled services.

What's left is to bring the people in. How do you do that? Provide the world with a free ticket to Disneyland… or in the Skype terminology – provide the world with free, high-quality and reliable voice calls, and viral adoption (of Skype? - No! of the eBay services platform!) is guaranteed (by the way, the topology of virus propogation is that of a scale-free network... like Skype's).

To conclude, by acquiring Skype eBay got:
- a proven, [potentially] infinitely scalable, reliable and cost-effective network, on top of which both Batch and Real-Time services can be deployed.
- a voice-enabled services platform - the future of the people's web 2.0
- a great viral service that will allow the creation of hundreds of millions of subscribers - ready-made prospects for the consumption of current and future eBay' services.

If that's not worth 4 Billion, what is?

This is the second post on Web2.0, SOA and the Future of Telecom. The next post will be dedicated to the Telco companies and their perspective on what's going on here.


By Blogger Paul Jardine, at 4:25 PM  

Why didn't they just license Joltid?? I heard that they didn't even get the IP for the P2P in the Skype purchase.
Are Joltid licenses $4.1 billion for 200 million users?

By Blogger Muli Koppel, at 5:23 PM  

Thanks for the question.

Skype is obviously not just peer-to-peer network. If it was just that then licensing Joltid would have been a logical step. But Skype provides the VoIP infrastructure and the viral service – so it's a completely different story than just p2p infrastructure.

There is, though, another answer to why 4B$ which is completely psychological. I think eBAY wanted to make it clear that it didn't buy a software company, but rather a Telecom company! When you look at Telecoms M&A, 4B$ is what I would consider an "average" deal. For instance, last month Vodafone acquired Telesim, a Turkish operator with ("just") 9 million subscribers for 4.5BUS$.

By Blogger Paul Jardine, at 2:57 PM  

Sorry, it took me a while to get back here. I'm not sure that I buy the 'not just peer-to-peer network' as Skype licenses its VoIP from GIPS, so essentially the part that is Skype is whatever authentication/authorisation is outside Joltid, the client, NAT traversal and the session initiation (that they developed instead of SIP).
However, I do agree that Skype could quite easily be valued the same as a telco, with the only qualification being that the ARPU of the Skype subscriber base is probably in the order of cents rather than dollars per month.
Skype is user-friendly and 'it just works' - that is the thing that really sets it apart right now. There is a significant advantage in being the market leader in the field, but I wonder how long it will last. Skype's innovation and momentum look like they might be drying up.
They face the dilemma that the telcos face, to retain control and place huge inertia on service development, or to open up and potentially give away huge amounts of revenue on the back of your network. 10% of X or 90% of Y? I believe I'd rather have 10% of X, but the telco position is, and will probably continue to be 90% of Y, even as Y tends to zero.
Ok, enough rambling! I'll connect to you on LinkedIn if that's ok.

Post a Comment

Wednesday, January 11, 2006

Web2.0, SOA and the Future of Telecom

I was researching several, unrelated narratives - one on SOA and Web.20, the other on VoIP and the future of Telecom - when suddenly the narratives got remixed into a surprising mashed-up synthesis, involving all these pro-tag-onists: Web2.0 and [the global] SOA; stupid networks vs. intelligent networks; Free Internet, two-tier Internet, and broken Internet; the forthcoming Telco War, [the inevitable] Google and so forth.

(* the more I use for my daily work, the more I feel like I'm living in a world of tags, where the tag is as real as the object it refers to. This is called Hyperreality – an interesting subject by itself).

In the post(s) that follow I will describe this synthesis, which claims that no war is needed and that the Telco companies are having a critical role in shaping the Read/Write Web, better known as Web2.0.

Web2.0 is the Global Point to Point

Dion Hinchcliffe, Dave Lintichum and others pointed out recently to the similarities between web2.0 and SOA: both enable easy reuse of existing APIs while promoting the creation of composite applications, or mash-ups. SOA is Enterprise-oriented, Web2.0 is WWW-oriented, but fundamentally Web 2.0 is the Global SOA, and SOA is the "mini-me" of Web2.0.
I tend to disagree. Hinchcliffe himself is explicitly referring to “significantly divergent” characteristics between the two: "SOA has much more central control, management, and governance while Web 2.0 is free wheeling, decentralized, grassroots, and with absolutely no command and control structure". Well, I think that the essence of SOA IS Management & Control, and that therefore SOA and Web2.0 are essentially different. Moreover, I see a total resemblance between the current state of the Read/Write Web and the pre-SOA (or the unmanaged) Enterprise. We used to call that architecture Point to Point, and web2.0 is nothing but a World-Wide Yarn Ball.

I already discussed the great error, IMHO, that Microsoft and IBM committed, by releasing the first WS-related standards without an adequate SOA management & control platform. By that they allowed rapid, unmanaged, application development, but with the same inevitable outcome of poor quality, availability, scalability etc.

The repetitive failures of Web 2.0 services, such as, and six apart – to name few that recently broke up – and the domino-effect they've created are the result of missing management architecture and infrastructure. Add to that the lack of identity management (the missing Identity 2.0) and the lack of any mechanism to globally control service abuse (*) and you receive a Read/Write Web that justifies the title “The Internet Is Broken”.

My argument is, therefore, that Web2.0 is in a desperate need for a Global SOA management and control layer, as the current architecture prohibits business-oriented, mission-critical consumption of mash-ups. And as I explained in my post on Always-On Services - the definition of mission-criticality is entirely subjective!.

If you accept this argument, then the inevitable next step would be to figure out who's capable of providing such a World Wide Management & Control Infrastructure. Asked differently, who's capable of managing the Internet?

My thesis, which will be presented in my next post, is that the Telecom companies are in an excellent position to be the Global SOA providers. This will require kind of an out-of-the-box thinking for the Telco guys, as currently they are busy preparing A WAR. ”We’re witnessing the beginnings of a titanic clash between the internet and the telecommunications industry”, says Alec Saunders in the beginning of his Voice 2.0 Manifesto. I believe this clash can be avoided, unless, of course, someone already paid for the battlefield.

Ambassador Trentino: I am willing to do anything to prevent this war.
Rufus T. Firefly: It's too late. I've already paid a month's rent on the battlefield.

The Marx Brothers, Duck Soup (1933)

(*) An astounding example of a service abuse is Hotmail. In an interesting article by ACMQueue, Phil Smoot, product unit manager in Microsoft’s MSN division, reveals that 75% of all mails are Spam. Handling spam before it gets to the service provider requires Intelligent Network filtering. Similarly to Cisco’s Network Admission Control which allows for first-line Virus detection and elimination already in the Network Layer, the Global SOA should allow for detection and prevention of spam or any other fraudulent/abusive usage of a service before it reaches the service provider itself. Well, at least as a first-line protection.


Post a Comment