Thursday, August 12, 2010

Smash the Management Silo … and bring me Web services!

Working backwards through a recent article in the technology portion of the July 31st, 2010, issue of The Economist, under the headline of “The end of Wintel”, I came to the closing statement “instead of being dominated by two monopolists (e.g. Microsoft and Intel), the market will be fought over by eight or nine more or less vertically integrated giants. Oracle, Cisco and IBM will vie for corporate customers; Apple and Google will scramble for individuals. IT, like the world, is becoming multipolar.”

Making sense of this I revisited two earlier comments: “firms are becoming more vertically integrated. For these new forms of computing to work well, the different layers must be closely intertwined.” Finding it hard to ignore the headlines about mergers and acquisitions this year, and the potential results too tempting to ignore, The Economist then points out how “the heavyweights in corporate IT are invading each others territory, too. That is the only way to grow, they believe. Also, clients love a one-stop-shop.”

But perhaps the biggest revelation comes close to the beginning of the article when the wise heads at The Economist sketch out how “processors (are) growing ever smaller and more powerful; internet and wireless connections keep speeding up …. Pushing computing into data centers (huge warehouses full of servers) and onto mobile devices … the shift to mobile computing and data centers (also known as “cloud computing”) has speeded up the “verticalization” of the IT industry.”

Where to begin with all of this – vertically integrated giants? Multipolar? One-stop-shop? Data centers (huge warehouses full of servers; also known as “cloud computing”)? I get the very real sense that even well-informed journalists are facing as many difficulties coming to terms with modern IT as do CIOs, and even data center managers. No, not all data centers are nothing more than a warehouse full of servers piled high up to form clouds! True, maintaining the “silo” approach of the past is becoming an expensive option but no, vertically integrated offerings stacked like layers of pancakes from IHOP and blindly pursued, are not a panacea or the immediate choice of today’s enterprise executives.

What is happening however, and is being driven in part by the significantly diminishing costs of computing power at a time when recessionary forces are driving companies to explore ways of doing a lot more with substantially less, very large enterprise-, or mainframe-class systems are struggling to remain relevant. While many in the industry champion big is better, collapsing the industry will only stifle competition and lead to price stabilization (i.e. a slowing in the rate of price reductions) – not a scenario many familiar with the needs to enterprise data centers cherish all that much, and hardly one conducive to helping them do more with less.

But is this what the future holds and is it really a better model? Where should IT be investing and what should be the priorities? On this subject I return to two very familiar territories – Web services and Data Center (and Application / Solution) Management. We really have mastered networking, for the most part, database software has become pretty mundane and mostly standardized, and what’s needed to secure data in flight and at rest is well understood, but before we head down the path toward vertical integration we need to come to terms with some of the approaches that make sense in terms of integration, as well as operations.

I tend to favor Web services when it comes to integrating a lot of applications spanning years of operations – the sum of the business logic wrapped up within these applications is just too large to ignore or to attempt to reproduce. Electing to go with a services model, whereby new applications can easily utilize and also can “expose” current business logic, just makes plain sense. Countless presentations have educated me on how easy it is to accept someone’s published WSDL and immediately become familiar with the data that will flow from the server(s) within the data center – it’s just that easy!

In the June 28th 2010 issue of InformationWeek under the heading of “Smash that Management Silo”, the author suggests that “when end users are unhappy with the performance of Web services and application, they tell you loud and clear. Calls and emails pile up as executives, line-of-business managers, and frontline workers express their anger over lost productivity.” Of course not every problem can be addressed by embracing Web services – brace yourselves for a tenfold increase in message sizes, for instance, perhaps more, but as a vehicle to help intertwine and integrate disparate solutions it’s way ahead of any other approach in terms of ease of comprehending by today’s raft of freshly minted college graduates! And the really smart product offerings available today offer more than one approach and just as often, complement true Web services offerings with solutions that rest less heavily on the server(s).

However, when it comes to management oversight of tomorrow’s vertically integrated solutions, even from the biggest vendors (who surprisingly, even today, show little enthusiasm for actually crossing over into the applications marketplace, no matter how tempting it may at first appear), it’s still a dog’s breakfast of fragmented solutions with poor insight down through all the layers, from the user interfaces to the bare metal components that make up a modern server. The InformationWeek article observed how “management tools specialized for each component in the organization’s array of applications, services, databases, networks, and systems scream that thresholds have been crossed. Running from one tool’s interface to another, IT professionals find it difficult to identify the root cause, leading to delays that can affect the bottom line.”

In the future, where Web services provide the glue that helps develop vertically integrated solutions on a scale The Economist suggests, and from warehouses full of severs indistinguishable from cloud computing models being widely discussed, progress will only be made that is commensurate with our ability to manage it all. As a warehouse of servers in a box and from a Pathway perspective, virtualized not too dissimilarly to what much of cloud computing addresses, the NonStop has supported all of this for years.

True, not fully standards-based, or supporting the modern programming models demanded of today’s CIOs, but that’s all in the past! ISVs in the NonStop marketplace fully understand the situation just as many of these same ISVs have been providing Web services for over a decade. And while The Economist may fail to include HP in the list of vertically integrated giants, let alone any oblique reference to NonStop in passing, I cannot leave behind the sense of déjà vu from it all. Perhaps it would be a lot easier if we could find a newer breed of journalists, more wise to the “cycles within IT”, that are a less anxious to explain data centers and more in tune with what it really takes to move the industry forward.

1 comment:

  1. Rather than looking for a new breed of IT journalists to bring more wisdom to the industry, how about looking for C-level managers not being overly happy with the current "state of the art" in IT ? I'd bet the latter ones are much more easy to find !

    Complexity is the main obstacle to a better IT. A better IT would come with less complexity and would provide better service levels at lower total cost of ownership. Yet, by lack of a holistic view on IT, decisions are delegated to lower levels of the organisation - which leads to suboptimal results. So, potentially some project cost can be reduced by acquiring cheaper hardware - but who calculates the increased operating cost resulting from such decisions, making the application much more expensive to run, for many years to come ? Not my budget, not my problem ...

    Seems that now some new thinking is emerging, and the offering of more integrated vertical stacks by major IT vendors like Oracle or IBM is a sign for this. After many years of mix-and-match, IT is becoming more proprietary again ...

    Is this good, or is this bad ? This will be measured by the results. It is quite likely that these vertical stacks will lead to less complexity and less systems integration at the customer level. Service levels and TCO are likely to benefit from this ...

    Looking into their own arsenal, HP might find that somewhere in some dusty corner they have a almost forgotten NonStop technology that fits quite nicely into that new scenario. It is also very well suited for the current cloud hype - after all, very special "clouds" run by specialized payments service providers are taking care for all those plastic money transactions since many years. That's a very good example for reliable and secure clouds that will be needed too - for sure, not all workloads are uncritical enough to be handled by plain vanilla clouds.

    Web services are a good means to have various applications talking to each other, also across organisational borders. But even without WSDL, there have always been ways to achieve this. Having the right strategies is more important than focussing on some particular technical implementation ...