SEARCHING FOR INSPIRATION in shaping the next computing architecture for IT, IBM Research apparently is looking inward -- literally. Taking a cue from the human body's autonomic nervous system, Big Blue has a vision of piecing together an autonomic computing network from across its far-flung development labs. As is the nervous system, the design of this new computing context will be flexible. It will respond to unpredictable events in intelligent ways, will be self-managing and self-healing by automatically correcting problems or fetching resources that can, and it will be always accessible so that a company's customers, partners, and suppliers can get to it easily.

   ADVERTISEMENT
  

Free IT resource

Hear how top CIOs turn change into a competitive advantage.

Sponsored by HP

Free IT resource

Attend the SOA Executive Forum: Breaking SOA Bottlenecks SOAExecForum.com/may2007

Sponsored by InfoWorld

RELATED LINKS

IDG ENTERPRISE NETWORK
• More App Development News...  (ComputerWorld)
• JBoss buys former HP middleware  (ComputerWorld)

TOP NEWS 


IT SOLUTION SEARCH
"Building an autonomic computing network is not optional; in fact, the future of the Internet and e-business depends on it," says Paul Horn, the director of IBM Research and caretaker of the company's autonomic vision. At the heart of this design is intelligent middleware that functions autonomically, according to Horn. He contends that a truly autonomic computing network will be an aggregation of autonomic networks, with that aggregation being completely transparent to users. "This must be an entirely heterogeneous environment, [one that is] platform agnostic, and one [that is] dependent on open standards. And middleware must support all modes of access to it, be they cell phones, kiosks, Web-connected appliances, embedded devices, or even PCs," Horn says.

If this intelligent middleware layer can be built, then IBM can deliver on its ultimate promise of autonomic computing, which is "e-sourcing," or the ability to purchase IT services in a "utilitylike manner," Horn says. Horn contends that in the future servers that make up this system will be more cellular and distributed, much like the human nervous system. That means we'll have systems and networks of single-chip cells that integrate processors, memory, and communications. These cellular architectures will make it practical to allow the computer to get its power out to where the data is, thereby eliminating some of the latency characteristics of current architectures.

IBM Research is currently working on software to exploit such an architecture. Horn describes one project, called Oceano, as a "scalable infrastructure for large-scale utility," that allows for multicustomer hosting on a collection of virtualized hardware. Via Oceano a group of servers can be automated to handle the IT needs of many users, including on-the-fly changes in the load requirements. Paul Horn sat down with InfoWorld Editor at Large Ed Scannell to discuss Oceano and a number of other topics.

InfoWorld: IBM Research has a vision of the future wrapped around this idea of autonomic computing. What is autonomic computing?

Horn: The body's autonomic nervous system does things like change your heart rate and breathing, which allows humans to physically adapt to any number of situations. This is very much the way you need to think about the middleware as it will serve the Internet. For us, e-commerce is about building technologies that enable businesses to reinvent themselves and their partnerships, so they can generate huge efficiencies and productivity. We call this Dynamic E-Business, which will be all about how you form your business relationships with suppliers and service providers and, ultimately, how to form virtual companies that come together to solve some need and then dissolve. All this will be done using tools and middleware over the Internet. E-commerce middleware that will provide value for us will start from our WebSphere Commerce Suites, which will become more and more self-optimizing, more automated. You will see how this ties in with the idea of the autonomic nervous system.

InfoWorld: How strategic is this idea of autonomic computing for IBM over the long term?

Horn: It is pretty strategic. The autonomic concept will encompass an enormous amount of what we do. It is a unifying theme across all of our research labs. It is an overarching vision we are using to drive what we do.

InfoWorld: How focused is IBM Research on middleware as part of this?

Horn: More and more of the value for IT will reside in what we call the intelligent infrastructure. In fact, the Internet itself is becoming more intelligent, and the thing that is giving it that intelligence is middleware. Middleware is doing things like transcoding, where you can change the data so it takes on the characteristics for whatever the end-user device is. It is also doing things like caching so that performance improves. And [it is being used for] things like directories, security, and QoS [quality of service]; it is all moving into middleware layers that are making the Internet more intelligent.

InfoWorld: How does your Oceano project tie in with this?

Horn: We actually have a prototype of this operating. I'll give you an example: Today, if we were doing a Victoria's Secret fashion show on a Web server, we would gather a bunch of servers, plug them in, and load them up with the [Web server] application. You would get the ability to deal with a lot of customers through brute force by just plugging in more servers. What Oceano can do is pull that server farm together automatically by sensing what the demand is going to be. It goes out and automatically brings down servers not being heavily used for, say, a sports Web site. It will clean off that site and bring those servers back up but running the Victoria's Secret application. It can do this without any human intervention. It can be a great productivity boost for IT. We think this is going to be pretty exciting technology.

InfoWorld: How do you bring this to market? Is it delivered as a standalone product or technology that is built into existing products?

Horn: There could be a few ways, and we have not decided how yet. It could be delivered as a software product. We could load it up on our servers to differentiate the way they operate compared to those of our competitors. We could bring it to market as part of our server business, or we could use it for our own utilities in our Global Services organization.

InfoWorld: How important are standards for completing this vision?

Horn: As a company we have been pushing these new [software] standards such as UDDI [Universal Description Discovery and Integration] and Sash that allow you to do high-level e-commerce in a platform-agnostic way. We believe they are critical so you can get business value out of information technology without having to hurdle all these conflicts. If you don't have standards [then] you don't know how to communicate between one marketplace and another or even between one business and another.

InfoWorld: What is IBM Research's role in helping IBM crystallize its e-sourcing strategy? And how does it evolve in the future?

Horn: For me e-sourcing is thinking about utilities, thinking about how you can provide value by keeping in mind that you are servicing numerous customers. A lot of that falls under Oceano, into having business-process standards so that from one source you can deal with multiple and very different businesses and different business applications.

InfoWorld: What role will natural languages play in all this?

Horn: Natural languages are probably one of our most significant focuses here. Look at where the data is being put online today. The amount being put into databases is huge and growing. But the amount of unstructured data is growing at an even more astronomical rate. I am talking about unstructured data like Web pages, HTML, video, and MP3 audio. That is all digital data that is not sitting in some database. Most people today need to find what they need on the Internet and get value from it. To do that you have to be able to search it, with more than just a keyword search on some Web page's search engine. You have to be able to find real, meaningful data by doing things like correlations and systematically understanding the various things that are going on with this unstructured data. It is probably the biggest way businesses will distinguish themselves in the future. So to get more value out of a text document, you need a natural language that can understand what that text is talking about.

InfoWorld: How soon before we see delivery of something such as that?

Horn: You can see the beginnings of it in something like [Lotus'] Raven. One of the things it does is you can work it to analyze your e-mail. As part of that analysis, it can determine if you are an expert on something. So your company [can] search for an in-house expert in a certain area by analyzing the text of your work. It is an expertise locator that essentially operates using natural language.

InfoWorld: Are there any other associated projects still cooking in the labs?

Horn: Yes, several. There is one [referred to internally as the Personal Speech Assistant] that is a PDA [personal digital assistant], a modified Palm Pilot. We have modified it for natural language and speech recognition. You can turn it on and off by speaking to it, manipulate the built-in calendar and address book, and surf the Web. It also translates English into several languages, including Japanese and German. It is only a research project now. We still have some work to do to improve its battery life, the translator, and vocabulary.