IBM puts WSDM behind IT automation

IBM puts WSDM behind IT automation

Summary: IBM has come out with a new package of services and software to promote greater automation of IT operations, and Web services is being employed throughout. The mechanism by which a lot of the inter-systems play that is required for automation is the WSDM standard, or Web Services Distributed Management.

SHARE:
TOPICS: Cloud
6

IBM has come out with a new package of services and software to promote greater automation of IT operations, and Web services is being employed throughout. The mechanism by which a lot of the inter-systems play that is required for automation is the WSDM standard, or Web Services Distributed Management.  WSDM was approved as an OASIS standard earlier this year.

I had an opportunity to chat earlier this week with Dave Bartlett, IBM’s vice president of autonomic computing, who said that WSDM provides the standard format for representing events across a network. "This enables us to take all the raw trace files from different applications, different vendors, different operating systems, networks, network, different databases, and converge them into a single format. You get an end-to-end view of what’s occurring in the system, and therefore very rapidly identify root cause of problems, and take automated action to fix the problems."

IBM has been working closely with HP and Computer Associates to promote the WSDM standard not only to enable autonomic computing, but also grid architectures, as explained in this article

Data center automation represents the next great frontier of computing.  Big Blue estimates companies can achieve a 30 to 50 percent time savings on IT tasks. IBM reports that it has woven more than 475 autonomic features into 75 of its products since 2001. Many of the autonomic features and technologies are designed to automate the process of locating infrastructure problems, which when done manually, can be very time consuming.

WSDM is actually a twofer, comprised of Management Using Web Services (MUWS) and Management of Web services (MOWS). That's because when it comes to management, Web services becomes both the problem and the solution. Not only do Web services need to be managed, but Web services themselves can serve as management tools.

If OASIS WSDM catches on, vendors will be employing MUWS within their management tool offerings. Thus, Web services could handle a variety of IT functions such as performance monitoring, load balancing, provisioning, security, versioning, and compliance. MOWS is a different animal, but may be our best hope for taming the spaghetti networks that could make Web services too untenable for SOA.

IBM has also submitted a self-configuration specification to OASIS, Bartlett says. "When you go to deploy applications, or any type of software, is it a standardized packaged format in which you can capture all of the dependencies that are required, so they can be handled in a programmatic fashion." 

Topic: Cloud

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

6 comments
Log in or register to join the discussion
  • Its like .NET

    Clone a copy of Java, and then remove the platform-independent piece! Uh, WHY? To corrupt it into being PROPRIETARY!

    SOA is web services, that get built up into applications that can EXIST ANYWHERE - you just connect to it to use it. Location independence means that ANYONE can offer the service - you just point at the server/service and go. This is arguably the BIG ADVANTAGE of using SOA - and why large computer companies hate it.

    So along comes IBM, and creates a SOA model that exists INSIDE an INTRANET. This loses the location independence and thus corrupts SOA into being PROPRIETARY. If I wanted to switch from IBM to using HP - I COULDN'T! What Stupid Demented Mother came up with this? ;)
    Roger Ramjet
    • Agreed

      Yes, an SOA or Web service in a silo is just as inefficient and costly as a proprietary app in a silo. A true Web service can be shared across the enterprise in a repeatable fashion.
      joemckendrick
  • SOA rhetoric and the Jenkins test

    Let us say, that a survey revealed that 98% of chief information officers are in favour of orientation towards service.

    Furthermore, a similar percentage believed they should have an architecture.

    The remaining two percent replied "our users are idiots, we implement what we like and when we like, with no thought for long term planning or architecture". This group are currently in conference with their chief executives discussing whether they would like to consider other opportunities or spending more time with their wives and families.

    We have therefore revealed massive support for service orientated architectures.

    Now we apply the Jenkins test. Roy Jenkins is a British politician and former cabinet minister. He proposed that you can tell if a politician's speech has any actual content by examining if any reasonable person could disagree with what they are saying.

    So if, for example, a politician's entire speech is about how he is in favour of democracy, freedom and prosperity then no one sensible would disagree (apart from a small, lunatic minority who wish to be tyrannised, enslaved and destitute). We can also agree that the speech could also be viewed as the mindless regurgitation of platitudes.

    On the other hand, if the speech discusses what form our democracy should take, what our freedoms are and how they can be preserved and how we can achieve prosperity (and for whom) then we have something we can disagree about and hence debate and therefore the speech has some meaningful content.

    Everyone agrees it is good to be service orientated, everyone agrees that architecture in systems (in the sense of a coherent plan) is good. The interesting debate is about how we achieve this. As far I can see 90% of discussion on SOA is empty rhetoric and verbiage. If you dig a bit deeper you find that all that is being proposed is a reiteration of the object orientated and hierarchical methods that are dubious in theory and have already largely failed in practice.
    jorwell
    • SOA is high on the hype cycle

      Every vendor out there now claims to support "SOA" in their products or services, which means a loss of credibility for the concept. The concept itself is somewhat nebulous, and we define it as a set of industry standard application components that support an end-to-end business process in an orchestrated fashion. But who knows when we've actually built something called an SOA? When do we cross that line from having a bunch of Web services to a functioning SOA? Just about every company has some type of Web service-enabled application up and running (meaning XML, SOAP, WSDL). A company may have 1,000 Web services for various point-to-point functions, but that does not an SOA make.

      SOA is a vision, or perhaps a roadmap, with a destination that probably will never be fully realized, just as democracy, freedom, and prosperity are such visions. But we need such vision and inspiration to give meaning and purpose to our day-to-day activities. That applies to politics, personal ambitions, business, and information technology.
      joemckendrick
      • It is the method's I am dubious about

        My experience with XML/OO based solutions makes me believe that the SOA approach is inflexible and code intensive.

        Also I think SOA looks like and attempt to paper over cracks rather than solve underlying problems.

        I am far more interested in approaches that integrate by allowing the definition of a logical relational model over the DBMSs of multiple vendors.

        This seems to me to offer numerous advantages.

        1. The data validation is in the database. The data is thus protected by whatever means you update it.

        2. The user or programmer can query and update data without recourse to procedural programming languages.

        3. The data of the business is a single logical model and should be viewed as such. If for historical reasons the data exists physically in different systems there is no need for users or programmers to know about this at all.

        4. Duplication of data can gradually be removed by substituting the duplicate tables with views based on proxy tables in the remote system.

        To me this approach seems straightforward and easy to understand because it is based on a strong set of mathematical principles, that is, the relational model of data.

        I have worked with object orientated techniques for many years and I am increasingly doubtful about their utility and the validity of their theoretical underpinnings. Every book I pick up about object orientation claims in the introduction claims that object orientation "reflects the way people think naturally". This is always offered without any references either to a philosophical justification for this position or to any experimental psychological evidence to support it.

        Even if this claim were true (which I hope it is clear I doubt) it would be a useless basis for constructing a computer system as a computer system is by its very nature formal and to construct one you have to learn to think in a different way from how people think naturally.

        Furthermore, there seems to be a basic confusion in the object orientated community between types and relations. In the relational model you use the same set of operators to manipulate any relation (table in SQLspeak). In the object orientated world you have to go through the tedious and unnecessary process of writing get and set routines for every damn element of every object. If I have a relation with two attributes customer_id and customer name this simply isn't a "type" of customer, it is a relation between the set of customers (identified by id) and the set of names. Why do object orientated enthusiasts find this so hard to understand? Aside from the logical flaws isn't it fairly obvious that the object orientated approach is extremely code intensive and application specific in comparison with the relational approach?

        I see XML at the moment jumping through all the same painful hoops that hierarchical DBMSs had to jump through in the 1970s in order to try and get round the lack of a proper theoretical model. This is why hierarchical DBMSs have been almost completely superceded by SQL DBMSs, that are based (albeit, far from perfectly) on the relational model. XML is the revival of a long and justifiably dead technology.
        jorwell
      • And a quick note on democracy

        I now live in a land where there are still many people alive who have experienced systems other than democracy.

        My observation is that people spend less time talking about democracy and more time just getting on with it than in countries that perhaps take their democratic rights more for granted.

        Consider, however, that I left a country that has no written constitution where fundamental rights are set out!
        jorwell