09.21.08
Posted in Green, Technology Market Trends at 3:55 pm by Tony Baer
Blame it on having been a history major undergrad, but it’s difficult to read the newspaper or track industry trends without having a sense of deja vu. The current inflation in energy prices is another such example -– in my lifetime, we’ve been through this treadmill a couple times before, during the gas shortages of after the 73 Arab-Israeli War, and the second run-up in energy prices after the Ayatollah took power in Iran.
Recalling our first real job out of college, as staff technical writer with a newly created federal research body, the Solar Energy Research Institute (now the National Renewable Energy Laboratory), we worked with scientists who were attempting to decipher how to commercialize junk silicon into solar cells; how to scale up wind power; whether ethanol could ever get the right carbon balance; and whether it made sense to build passive solar houses that, as we discovered personally, are fine to live in as long as you don’t occupy them during the day (as the building stores heat during daylight hours, they get pretty hot). And we worked with staff economists who examined disconnects in tax and corporate accounting systems that discouraged the kinds of long-term investments that it would take to make the economy energy efficient.
So excuse us if we’re a bit jaded to all the concern over energy efficiency –- we’ve been through the drill several times before, and each time the economy subsequently adapted to higher prices as demand remained dwarfed by artificially-manipulated supplies.
But this time things are different — even if some of our smartest, conservative-leaning friends have yet to be convinced of whether there’s real proof of global warming. (We’re tempted to say that the omnipresence of Rush Limbaugh and similar dittohead windbags are ample proof that there’s lots more hot air circulating around the atmosphere than ever.)
The big difference this go round, besides awareness of the problem of global warming — are structural changes to the world economy that will guarantee that this will be no repeat of 1973 or 1979. Specifically, we’re talking about the entry of the world’s two most populated countries -– China and India -– to the world economy. And that’s just the tip of the rapidly-vanishing iceberg, as other lesser developer regions are also ramping up. The bottom line is the huge expansion of the global middle class on levels that dwarf even the levels of growth of the age of industrialization and the post World War II period.
Translated to resource consumption, it means that this time round, we won’t be able to simply adapt to higher gas or commodity prices because they will become moving targets as increased global demand becomes less elastic. Recession in the short term will only moderate what will become an ongoing upward trend. To make economic growth sustainable -– not just in the U.S. and developed economies, but across the globe -– resource consumption must become more sustainable.
And back to global warming, let’s add the fact that we will have to deal with CO2 at the source. There aren’t any easy ways out. For instance, clean coal may sound nice, but nobody’s developed a sustainable, economical approach to figuring out how to sequester all the CO2 that will spew out. Nuclear? There’s still the issue of long-term waste whose half-lives extend to geologic time.
There are numerous hurdles to sustainability, which we won’t even try to attack here, such as the lack of sufficient grid interconnection that could enable dispersed generation of, say, wind power.
What reminded us of all this was a recent note from enterprise software industry veteran and friend MR Rangawsami, pointing us to a recent speech by Google CEO Eric Schmidt, which besides providing some clever uses of Google Earth, summarized well the challenges to making the transformation to energy sustainability –- and a number of common sense solutions for addressing it. Just one example was how to deal with the problem of dealing with peak energy consumption (times such as 4:30 pm on hot July afternoons when every air conditioner is running full blast). That’s the point where utilities must fire up their most expensive generating plants and consume the most expensive, and usually imported, fuels. Schmidt described it as akin to solving a classic distributed computing problem where you enable lots of small generators or such as plug-in electric car batteries or rooftop solar cells (the hottest days of the year which typically have the highest consumption also tend to get the most sunshine) become part of the network so you can shave those peaks.
That’s just one example -– if you have 45 minutes to spare, check out his talk on YouTube, and if you dig Google Earth, click on the high resolution viewing option.
Permalink
09.15.08
Posted in Application Development, BPM, Business Intelligence, Data Management, Database, Middleware, SOA & Web Services at 12:04 pm by Tony Baer
With the vast majority of the world’s transaction processing occurring on IBM platforms, it’s little surprise that IBM is no stranger to complex or business event processing.
Last week, IBM convened a press and analyst briefing to raise its profile in the nascent market of complex event processing. Of course whether it’s a real market — we’ll deal with that one later.
IBM claims to have the largest base of business event processing customers, which it estimates at nearly 3800 worldwide. The catch of course is that the category is entirely subjective depending on where you set the boundaries. It could be an application that is triggered by a discrete occurrence that requires an action, or one that is triggered by a complex series of events or detection of a trend from a series of occurrences. There are event processing applications that require extremely high rates of throughout and minimum latencies, while other event processes have far less stringent performance requirements. The use cases are all over the map.
In turn there are many ways to detect, process, and respond to events ranging from adding time-based parameters to SQL commands to intercepting streams of data in cache and an d applying sets of rules that define what constitutes an event, and what events merit proactive response. The variations start when you think about application of rules-based processing, business intelligence analytics and dashboards, process-based orchestration, and more.
Not surprisingly, IBM has had a range of technologies in its various software products over the years that handle events that are complex and/or involve huge torrents of data. It has business activity monitors; high performance, low-latency messaging systems and in-memory databases; event detection; and predictive processing algorithms. And of course, it has CICS, the nearly 40-year old mother of all transaction processing systems, which processes 30 billion transactions (obviously, not all of them high performance) daily.
To that it has recently augmented the portfolio with higher-performance versions of WebSphere and made several pivotal acquisitions including AptSoft, which offers a declarative development environment for business users to define events and how they should be processed; SolidDB, which provides an in-memory database; Infodyne, targeted at capital markets and hedge funds for configuring financial trading events; plus Cognos, which provides dashboards and BI analytics and Ilog, for defining complex rules processing. And for its information management business, IBM has just announced InfoSphere InfoStream processing, a new stream processing engineer that performs in memory parsing and manipulation of event streams for extremely high performance, low-latency scenarios, such as financial markets where the technology was betas tested.
Clearly, IBM has plenty in its armada to offer for a wide variety of event processing use cases. With a number of rivals which focus on specific pieces, such as time-based SQL or event stream processing, IBM arguably offers the widest, although not necessarily the most mature portfolio in all event-processing areas. Admittedly, in certain product areas, such as stream processing, IBM’s rivals such as Progress Apama or Tibco have more mature product that are focused on specific niches. In time-based SQL, you have startups like Coral8, Streambase, Agent Logic and others that emphasize a SQL-like approach with IDEs that appeal to the same developers who have developed data-driven applications. With rivals talking complex event processing, IBM decided that it was time to raise its profile to say that they’ve been performing the task for years.
The dilemma is that event processing is not something you buy with a single shrink-wrapped application. The technologies are so diverse, and so are the use cases. For starters, lets ask, just what is event processing? Some call it event processing to refer to something generic; there is also complex event processing where you parse intricate sequences of events or occurrences that could only be detectable by machine; and then there is business event processing, which is the idea that we’re not just talking about events in the abstract.
Evidently, we weren’t alone in trying to navigate the confusion. As Neil Ward-Dutton stated, IBM’s label, Business Event Processing, refers more to how you are processing the data rather than the business rationale for what you are processing and why you are processing it.
Whatever you call it, it refers to an extremely wide range of subjectively categorized use cases that include any combination of the following:
• High performance, low latency
• High throughout
• Structured and/or unstructured text or rich media data;
• Detection of highly complex event patterns requiring complex rules processing
• Parsing of streams in-memory
• Adding time-based constants to SQL processing
• Providing declarative event processing design environments or configurations for business/non-IT users
IBM is only now realizing all of the pieces that it has, and is now beginning to identify possibilities for reuse of its event processing portfolio across its information Management, WebSphere, and Tivoli brands, not to mention logical technology pairings, such as adding Cognos business performance management solutions in conjunction with its various event identification technologies and vertical industry SOA frameworks.
As we noted before, we wonder whether there is such a thing as an event, business event, or complex event processing market per se. We have our doubts, not because the category is so difficult to bound, but because event processing is not an end in itself or discrete market. Instead, event processing is part of business solutions ranging from telco provisioning to capital markets trading, check or credit clearing, supply chain optimization, law enforcement and homeland security, and so on.
IBM’s challenge is not simply to add the technology stacks to one another but to devise a framework of technology piece parts, implementation patterns and best practices that would provide a solution focus to the reference architecture that it has already begun piecing together.
Permalink
09.10.08
Posted in .NET, Application Development, Java, Standards Development at 4:12 pm by Tony Baer
Chalk another ancient family feud to evaporate. Microsoft and OMG have agreed to bury the hatchet. The announcement was carried in a terse press release backed by a stilted, condescending video of server & tools VP Bob Muglia interviewed by a junior PR or marketing aide who was obviously reading her lines off a teleprompter.
Microsoft and OMG had been butting heads back to the dawning of client/server, as each backed their own definitions of what would define the component architecture for the next generation of object-oriented software. Microsoft had its COM model, intended to provide its answer for enterprise Windows development, vs. CORBA, the model promoted by the rest of the industry (read: UNIX and legacy server providers) as the multi-vendor alternative. In other words, Microsoft’s de facto vs. the rest of the industry’s “open” standard.
COM and CORBA eventually battled to a draw; COM was too platform-limited while CORBA was too complex. Both eventually ceded ground to more “modern” architectures: .NET framework, which provided a multi-language run time requiring managed (e.g., strongly typed) code); and J2EE, which provided a more accessible answer to CORBA. In turn, .NET and J2EE had their share of struggles: Microsoft waging a multi-year ground war to eventually get VB developers to adopted more disciplined development under .NET, and J2EE remaining too complex until open source alternatives such as Spring and Hibernate forced the JCP’s hand with the lighter weight EJB 3.
We had a feeling that something was afoot when at TechEd, Bill Gates’ valedictory keynote endorsed UML for its emerging Oslo business process initiative.
Ironically, Microsoft was one of the original backers of UML, and – justifiably in our view – backed away when UML itself grew too complex in its 2.0 version. To recap, UML is a visual modeling notation for describing the logical architecture of a software program. The problem was that with version 2.0, UML got larded with additional diagrams for modeling physical deployment of a program, a development that unfortunately made model-driven development (an OMG initiative) look far too complex for ordinary mortals. While some architects simply stuck to UML 1.x diagrams, others such as Microsoft began looking for more intuitive domain-specific language (DSL) alternatives.
Microsoft is still working with groups like the Business Process Alliance regarding DSLs. But more to the point, it realized that at the end of the day, UML remained the de facto standard of model-driven software development. Rather than start from scratch, Microsoft figured that the most direct path to Oslo, which is based on the notion of model-driven development, had to run through UML.
Consequently, Microsoft vs. OMG was just one of those wars that just had to end with a whimper.
Permalink
09.03.08
Posted in Cloud, OS/Platforms, Open Source, Technology Market Trends at 4:28 pm by Tony Baer
Firing the first shot that tells you the summer’s over, Google yesterday unveiled Chrome, their skunk works project for developing their own browser. Given the dynamics of the browser space (it’s not a market, but a means to an end, which is controlling the gateway to the web), it’s not surprising that reaction can be summarized as follows:
1. It’s part of Google’s grand plan to challenge Microsoft by adding the last piece to what would be Google’s enterprise desktop, app dev platform, and cloud.
2. It clouds the waters given that Google just extended its sponsorship of the Mozilla Foundation for 3 more years.
3. Chrome is ultimately intended more for the kind of “power” browsing that would be required for the enterprise desk or webtop. The obvious goodie here is completely independent tabbed browsing, where each tab is its own session – meaning one tab crashing won’t bring all the others down. It’s the kind of feature that came to Windows beginning with NT and Windows 2000, where a single window did not always have to crash the entire client session and it’s about time that the Internet experience become similarly robust. This obviously oversimplifies all the possible wish lists for features that could improve the browsing experience, with security being the obvious piece, but more robust tabbed browsing is an obvious missing piece.
4. Chrome originated because Google realized it had to own the entire stack and optimize the browser for the Google desktop if it were to present a viable alternative to Microsoft.
5. Google extended its Mozilla partnership because it couldn’t suddenly pull the plug and transition to a technology that is barely in alpha phase. Open source blogger Dana Blankenhorn contends both are complementary; that that Google will ultimately push a dual tiered strategy, pushing Firefox to consumers and Chrome at the enterprise.
Regardless of your take, keep in mind that whatever Google’s ulterior motives, consider the source. Google, much like Microsoft, is so gargantuan and has so much resource that its product, technology, and business development strategy is highly decentralized. The typical scenario is that there are multiple groups vying for development of the next great thing, and that the one with the best technology, market plan, and/or political skills typically wins out. In large part that’s how one can explain inconsistencies in Microsoft’s strategies, as recently revealed with Oslo, where a new workflow engine was developed in competition with existing BizTalk Server. So we’re not surprised that the group working to optimize delivery of Google Desktop on Firefox is different from the one hatching Chrome.
But our “aha” moment came when we recalled last week’s unveiling by Mozilla of its own Alpha, in this case, a natural language text search facility in Firefox that it code-named Ubiquity. In other words, Firefox was also treading on Google’s doorstep. So now you have a case of the ubiquitous search and advertising engine that is striving to become the ubiquitous enterprise webtop and compute cloud with a market cap of nearly $145 billion, and a tax-exempt not-for-profit corporation that racked up maybe $6 million sales in all of 2007 that has a respectable but hardly universal market presence, and the answer is obvious: Firefox is clearly about the consumer while Google’s dead serious about the enterprise. Or as Blankenhorn stated it in a prescient post filed just prior to the Chrome announcement, there are two Internets –- the locked down one in the office, which probably restricts you to the Microsoft IE browser, and the home Internet, where you can use Firefox or something else.
Our take is that Chrome’s prime target is replacing IE in the corporate Internet, leaving the home one as table scraps up for grabs. Our sense is that is where Firefox’s ubiquity is headed – if some third party mashed up that capability using a more graphical metaphor, it could provide a key enabler to monetizing the mobile web. But that’s another story.
Permalink