The Future of Web Services April 25, 2003 - O'Reilly Emerging Technology Conference L. F. (Felipe)ĘCabrera, Ph.D. [Berkeley] Microsoft, Inc. Software Arch at MSFT regarding XML. [no link in research.microsoft.com] Felipe is a self proclaimed geek C# hacker "Web services architect" Doesn't want the question "Is 'the future of web services' a promise or a threat?", yet. So far, each Microsoft presenter has been a comedian about world domination. They know they are pilgrims in an unholy land ;) Today: web services are a universal app interconnection fabric 17 specs for routing, security... WSE 1.0 for .NET WS-I is operational [heh] Moving forward: make it easy move to distribued applications distributed apps reflect business needs move to loosely coupled bus relationships means that bus will reply more on interconnectivity Talking about mergers where CEOs tell techs to "just make it work together!" "software wants to talk to software" - I didn't know software 'wanted' anything. [DK] About his slides, apolgetically, "I go to the ... and it is like a transaction, I trade words for slides." [graph] impact of increazesd competetision...average lifetime in s&po 500...the lifetime of a big company has been going down. I suceed, get to be big, and smack die fast. To reverse the trend you need software...for every problem in the world throw software at it. Throw software to the problem...that is all we do...we have one hammer...everything else looks like a nail 'For every malaise in the world, throw software at it - Thats MSFT strategy' Why do you sell the Xbox .... Too many intelligent people at MSFT (smirk). [PS] ... random thoughts on unix, berkeley tcp/ip .... try to be 'wire neutral' (=tcp/ip ??, whats the point) Who is Don Chamberlain ? (circa 1983) - Researcher at Almaden. Wrote an appendix to a technical paper, which was SQL, then created XML infrastructure stuff. The infoset = http://www.w3.org/TR/xml-infoset/ 'We will fail if we can't get interop' (WE=MSFT or WE=everybody?) WS-I toolkit where is it now ? Ahh, very happy that everyone overinvested in wiring the planet (a la Gilder) Good point ! So how do we exploit all the unlit stuff ... (implication web services makes network app development easier so we'll see more network apps?) Early adopters are building at great cost some web services ... So why the stack of specs ? Because dist computing has many notions ... so you need specs for each notion analogous to the specs for threads for lightbulb, initially Interorganization web services retailer -> (manufacturer -> purchase order) -> supplier Notes on Berkeley: We put out TCP/IP stacks, but we liked Xerox's stack. DARPA paid the bills, so we used our stacks. That's the reality of it. Required Capabilities - Example P.O. message level security routing ... missed it... Acronym Soup - XML, XSD - XML (DTD), WSDL, SOAP, UDDI, HTTP, XSLT, PSVI, DIME, MIME Very proud to not have been involved with any of the soup If you slept for a few years, you wouldn't know any of them Innovation happened! hehehe Some of us take a slice of inovation and feel pretty good, but then innovation happens and we are behind again... D"The trend is toon Cadd methaada "mbtae relainn, IBMhancing restehe arcprhr e-existing Web Services infrastructure." "AND HE LOOKS PRETTY NORMAL, FOR A GUY WHO GAVE SQL TO THE WORLD" ABOUT DON CHAMBERLAIN, who created SQL Infoset - a "very cool concept" that MSFT really likes and wishes they had understood it years ago thinking about message passing as the paradigm brought up to demonstrate that the network is evolving His group have been trying to be disciplined: modular and composable factored to stand alone or work together general purpose agnostic to place or origin standards based multi-vendor interoperation is critical federated no central administration, control, failure "Hey man, just open a socket. Come on, open a socket..." MSFT is trying hard to have an architecture to scale up touts WS-I 160 people encourages interpretation Tim O'Reilly - Q: My question with the big web services stacks is that most innovation comes from low barrier entry, and a stack nine miles long doesn't make it easy. A: There are building blocks that are there. We think that IP was given by god, we all love the wires that bankrupt companies wired the planet. Talk about barrier of entry.. Early adopters are building (at great cost) the first wave o [slide]f web services . Want to make that easier. O'Reilly: But there were more complex competitors to IP that lost out. A: use what you need Q: MSFT doesn't have a tradition of open standards, how do we believe that MSFT will interoperate in the future A: MSFT has been hiring more people to work on standards than lawyers, these days. Hiring more people to work on standards than lawyers! Q: If the GPL is a kind of guarentee for certain rights and access to code, what is the equivalent doc for MS? A: "What is GPL" (he must be kidding that he doesn't know what GPL is) [he totally knows] But his response was to call it a legal issue and so it is outside of his area. But htis ignored the 'msft said the same of java...' Where is the commitment to standards? Q: Your response to Tim was that "The task is complex, so the solution is complex" but there are other ways to do this, eg. Amazon's solution A: I agree. We need all of the pieces, and people should use what they need and nothing else. Hardware store metaphore. My job is to provide the building blocks. Section 5 of the SOAP spec, encoding. What were they smoking? So, MSFT is deprecating it at great cost and interoperability. you guys go do the technical stuff, and don't mind the effin standards. we will just deprecate section 5 of the soap standard,if you don't need it, just don't use it. (my interpretation) Q: Rather than discussion the standards specifically, how about describing how MSFT tech will use them and how that would affect other people's software? A: MSFT, partners for many of the specifications, commitment to show that the software interoperate with partners MSFT .NET security works perfectly with WebSphere Will demo with all partner platforms that they interoperate at release Just announced 2003, three years to release a platform is too long, so they have WSD to release in 6-8 month cycle [yetch] Q: As far as interoperability, it seems that one way to attack interop is to have these layered specs, but in HW unrelated members get together to show interoperation A: Yes, interoperation drives us crazy. People get together and test out their software against each other. "Whoever wants to play in this, we help." From the audience: some problems of interoperability come from XML schema committee simple models don't provide an asychronous pattern, for which there is a business need Another Q: Would you say something about trust model establishment (esp for non-.NET services) A: There is a roadmap for id services, cert services, access auth services (all pluggable) and these help establish trust relationships. Q: The approach is to develop specs and validation suites, which is an old approach. Another approach would be to make a reference implementation open. A: All interoperability tools and test suites will be open. Q: Not the test suites, the implementation A: These specs are to the point, and the length is all required. So if they are long, it's because the problem is complex. A: The problems haven't been the formats, but the semantics Q: Test suites and app samples will shake these out A: It is hotly debated by executives, but I am just a poor developer I feel that we're building the future solid code serious people e.g. IBM bigwig helping spec and build transaction model We have time for a couple more tomatos Q: Can you suggest how MSFT will eat its own dogfood? A: Refs case studies microsoft.com/casestudies Q: Is MSFT betting the farm A: Absolutely Anecdote: my project was a skunkworks started by bill and balmer (to protect the project from instant death) ... for someone who has "never played with Clippy", he hasn't found a way to turn it off yet WS-I baseline standards - XML 1 - SOAP 1.1as like 17 and you need a spec per notion. I 2.0 - HTTP 1.1 ---- metadata ---- Observation from the peanut gallery: [8:52] TomCoates: Ok - someone has to tell this guy [8:52] TomCoates: that his marketing departmen JamesDavidDuncan observed that tht [8:52] TomCoates: when they make slides [8:52] TomCoates: keep using little nice bits of Apple OSX iconography [hee hee hee :]e Macromedia slides had this too. MSFT is so huge, that the open folk are in a huge battle (meaning HUGE) with the closed folks. There are so many people working there that the social effects of the battle are complex It sucks that the lag has made interleaving a problem with editing, esp when more people are on. I'll email everyone on the list with the text. Append Email addresses for bounceback: esinclai@pobox.com richard_gayle@excite.com brian "we are at a stage where you have to protect the good ideas before the good ideas get killed.".theodore@reuters.com tfsmith@parc.com allen@hutchison.org dan@inmyexperience.com jmay@pobox.com acevedo@well.com pshah@sxip.com rich@testingrange.com etcon@crystalflame.net gary_ora@beagledreams.com mgraham@mail.ivillage.com dan@wde.com Append links and references: Graph of Corporate Lifetime is on page 7 of this pdf from the book "Creative Destruction" ( http://www.mckinsey.com/knowledge/books/pdf/CDch1.pdf ) A discussion of the Book "Creative Destruction" ( http://www.mckinsey.com/knowledge/books/book_creativedestruction.asp )