The network services ecology

The network services ecology

Summary: After a two week hiatus, Sun CTO Greg Papadopoulos (and former MIT computer science professor) resurfaced with another blog item on a familiar Sun theme: utility computing.

TOPICS: Oracle

After a two week hiatus, Sun CTO Greg Papadopoulos (and former MIT computer science professor) resurfaced with another blog item on a familiar Sun theme: utility computing. Greg expands the vision a bit with a diagram

Topic: Oracle

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


1 comment
Log in or register to join the discussion
  • Perspective vs. Objective


    You call that a diagram... okay, maybe it's best we don't go there.

    On a more serious note, you appear to be saying that talk of a new balance of power in the computing industry is over-done. I am inclined to agree with that basic premise - but not with your rationale, timetable, or your views on complexity.

    You claim that " ... refactoring of enterprise IT is at least a decade long, industrywide effort, and today requires synchronization and integration of thousands of disparate moving parts and standards development ... ". This is partly true, there are many parts and many new standards are required and, partly, not true - it does not require the support of the entire IT industry nor will it take ten years. IT customers (banks, retailers, manfacturers, shipping, governments and so on) only need to agree among themselves (i.e. a large number of banks leading the pack will change all banking) what standards they need for a major shift in the balance of power between IT suppliers - and between the buyers and sellers of IT in any particular sector.

    They will not need to "bulldoze most of what they have and start from a clean slate" for two fundamental reasons:
    - The economies of scale of outsourced 'old' technology will enable investment ahead of the curve; and
    - In commercial markets there is a 4 to 5 year replacement cycle (i.e. write-off) for technology in any case.

    That last point might seem likely to exclude government, but don't be so sure - once the commercial sectors have demonstrated feasibility...

    Neither will IT customers need "the most modern, not fully tested technologies, primarily packaged together and developed by a single vendor" - because:
    - Buyers are planning to give this problem to service suppliers who will provide guaranteed interoperable infrastructure; and
    - Buyers are writing the standards for process, infrastructure, and quality of service now, with vendors.

    Buyers are immunized against unmanageable systems by service provision if for no other reason than it is the service provider's problem. Yes, mistakes will be made but, within three years, it will be possible to have your service contract underwritten by a standard business continuity policy, so standardized will the service infrastructure become (for the outsourced/service-provided 80% of IT that is admin.). You could have argued that most buyers, and suppliers, are not yet skilful enough at this wrinkle - and I would have agreed with you. However, you appear to hold a much stronger view than that, and I can't help wondering why?

    I can only speak from my own experience. In the last eight years I have only advised IT suppliers in the Telecomunications, Automotive, Medical, Government, Finance and Energy sectors, so I cannot claim to have your broad perspective. However, in that time, I have concluded that many sectors - indeed most - are more than ready to make the switch between the traditional IT governance model:
    - Buy vs. Build;
    to the new IT governace model:
    - Service vs. In-House.

    Yes, they still have some things to learn about specifying and managing service contracts (in which I include outsourcing), but it is clear to me that most know enough to start migrating between those two models.

    You went on to claim that; "the price of software and infrastructure as a service is directly related to the complexities that service providers must endure to deliver services". That is certainly something that large, existing, service providers will tell you - and, to be fair, is partly true because not enough of their customers buy service-over-product contracts from them (to help them lever good economies of scale and scope) - yet. My read on that is that standards, and the availability of open source alternative components, are major determining factors in creating the commercial market for services. In many sectors, both these things will come this year. In some markets (finance springs to mind) this will also be in a very mature form.

    It could be that the major suppliers will succeed in putting open source back in it's box - in which case your ten year timetable may turn out to be conservative... However, even if the European Union eventually decides on a US-style software patent regime (I pray every night that this will not be so...), I am not inclined to believe that the OS community will fold up it's tent and steal away into the night.

    I don't mean to be rude Dan, but your comments on selling CPU cycles vs. enterprise applications as network services seem to me to be misinformed. That is the view of a dyed-in-the-wool, buy vs. build, CIO who is so long in the tooth he started out as a Data Processing Manager - which is to say: a common view. It is also a rapidly dying view. In many instances CEOs and Exec. Chairmen are employing a new marzipan-layer of consultants to talk business to the IT suppliers. They are re-drawing the IT supply map and Greg Papadopoulos' comments merely reflect the dialogue he is probably being forced to use - by his most senior customers. I can say that with confidence, even though I don't know anyone at Sun. The bottom line here is exactly that - show me the money. Major IT suppliers are falling over themselves (never mind each other) to talk turkey to many businesses because they know that service standards are within an ace of changing many industries' IT. They know this because, for the last five years, they have been sitting round standards tables with those same customers answering the question; "Bill, Scott, you can do it that way, yeah?" with a nod of the head...

    Yes this industry likes grand-sounding talk of revolutions (yawn), and this has been a long way to say; Not ten years, will you accept five, maybe?

    The world does not like revolutions, so evolution will be the order of the day (or of the next five years), but even so it will be a big, rapid, and fundamental change.

    You asked about winners and losers. My advice is look at a couple of companies that are
    - Long in this business;
    - Have very broad portfolios; and therefore
    - Demonstrate the power struggle every day.

    IBM, Oracle and Hewlett Packard are the easiest to read, but there are others. Easiest of all is IBM who's portfolio looks something like this (poetic license taken):

    - Consultants;
    - Sevice Providers;
    - Outsourcers;
    - Managed Infrastructure (e.g. on-site mainframe personnel, server farm designers) Providers;
    - Post-Sales Service Contracts personnel;
    - Sell-and-Support (a.k.a. Sell-and-Forget);
    - Retail Sales (direct and indirect sales of 'systems', e.g. Storage); and
    - Wholesale (to other badge suppliers, incl. parts supply).

    Almost every day IBM will make public some move, some policy change, a supply deal, a re-org or an appointment. But put them together and they don't always add up. No, I'm not going to make it easy for you - go and look for yourself! IBM is a company that is simply trying to understand the same future as you identified - no criticism of their, often excellent, managers is implied. That said, they are clearly struggling to shape a company that can do everything in IT. They have made some of the obvious decisions - hardware, except at the very top end, is already a highly commoditised business and margins are likely to be squeezed for the foreseeable future. Goodbye PC business. But what of tough choices, like systems management software? On the face of it Tivoli for On-Demand business is a great synergy. But look at it from the view of, say, Accenture - assume they are an IBM Tivoli customer. In the short term (as IT buyers struggle to come to terms with the fact that they don't specify the hardware and software any more) they will, probably, see a need for having some in-house Tivoli expertise.

    But what of the longer term? As the standards take hold and customers increasingly request service contracts that specify only: transparency for management oversight, business functions, standards compliance, and quality of service guarantees (incl. business continuity) - a specific management suite is of little importance to the sale.

    Now go back to IBM. At what point will competition for Tivoli, and our need for maximum flexibility in our supply chain at our service oriented level (i.e. top level, high margin, business) cross? What does that mean for Tivoli?

    It is important to remember that I used Tivoli only as an example. In reality Tivoli probably has a long future with IBM, if only because internal transfer prices will win the high level deals and will maintain consolidated margins. Even so, by now you've probably guessed that my rough-and-ready table of IBM capabilities is (very roughly) how the industry will define winners and losers in the service era. The more service-oriented, business-oriented, and solution-integrated you are the more likely it is that you will be the piper, and that you will call the tunes from all those lower in the stack.
    Stephen Wheeler