Are We Entering a Golden Age of IT?

How Pioneering Technology Architects View the Current Business Challenges and IT Opportunities

For over 10 years, our small band of technology architects (many of whom have now turned into business strategists) has been meeting twice a year. This is my “by invitation only” group known as Patty’s Pioneers. Each year, I learn an incredible amount from the rich sharing of problems and approaches these guys and gals are using to design and evolve their businesses’ technology architectures. The same patterns re-emerge over and over again, across industries as diverse as financial services, ocean shipping, energy management, retail, information services, manufacturing, supply chain logistics, and solving the world’s largest problems.

Each year, the patterns of our conversations form a semi-coherent picture in my head. I thought you might be interested in my non-Geek’s view of the way I see that these brilliant and persistent “big thinkers” think about the things they work on. (Pioneers: I have kept the quotes anonymous, but I am happy to attribute them! My apologies to any of you who may feel I’ve missed the point or over-simplified! Please chime in!)

What’s Going On in the Economy. Many of the Pioneers have deep financial backgrounds. So, of course, we spent some time talking about the state of the economy. I think the consensus was summed up best this way: “There was a fraudulent set of activities that started the butterfly effect of mistrust. Each leg of trust got knocked out. …Trust will come back when there are profits to be made. Everyone’s going to wake up and want to make money.” We had a really interesting and informative discussion about economic trends in different parts of the world, which included the predictions that European businesses are in a lot more trouble than most people think, that most small companies that are importers and exporters are going to go out of business because they can’t get letters of credit, and that the stock market will “bottom out” at 5,000 to 6,000.

What we can do about it? Pioneers felt that what they can do to help is to provide more transparency. All of the systems and architectures they design have transparency built in. Everything is visible and understandable. Proprietary stuff may be hidden behind APIs, but it’s not impenetrable. They also pointed out that the companies that will do best are those in which the top executives understand the means of production in their industries. Executives who are good at “managing” are no longer able to cope. Executives who have a visceral, emotional feel for the means of production in their respective industries will survive.

Why We’re Entering a Golden Age for IT. Pioneers are convinced that tight times call for innovative uses of technology. They see a golden age of IT when those who have already invested in smart architectures will vault ahead of those who have under-invested and/or invested in the wrong approaches. “Our board recognizes that they’ll need more information that’s processed in ways that are digestible to them, ways that let them better synthesize that information.”

What’s the Right Way to Think about Technology Architecture?

Model the Business. Most pioneers create an abstract model of their business using business objects with attributes and/or metadata. This is a living model. It evolves organically to reflect the dynamic relationships among entities and the attributes that become important over time. “Information is the proxy for reality. If you don’t have a proxy for reality, you’re doomed!” The brilliance in the systems that Pioneers create is that they enable the people in their businesses to understand how their businesses actually work!

Monitor Events and Exceptions. While traditional business process consultants and IT professionals think in terms of transforming, streamlining, and automating workflows, Pioneers view the world as “one big exception engine.” Stuff happens. Automated processes are only as good as the exception handling they provide. Pioneers assume that exceptions are the rule. So everything they design assumes that events will be delayed, that people will change their minds, that interruptions will take place.

Combine Objects and Methods Dynamically Each Time that a Service Is Requested. Pioneers favor near real-time binding of behaviors or methods to objects or entities. For example, a “check credit” service may be using different third-party credit rating services each time that request is made, depending on the context — Whose credit are we checking? For what purpose? Which ratings services are most appropriate right now? How many do we need to check? In which order?

Since change happens and improvisations are the norm, it makes the most sense to instantiate the state of the virtual view of the real world as close to the time that something (transaction, state change, event trigger) happens as possible. That way you can ensure that you always have the right context for the situation at hand.

Maintain the Current Version of the Truth in Cache/Memory. In order to ensure that their dynamically binding applications perform well, Pioneers also make lavish use of memory and disk cache to store the current state of everything. That way, the only time there’s a performance hit is for those few things whose state may have changed since the last time they were called into play.

Analytics and Computation. Pioneers design and use many algorithms that are computationally-intensive and which require aggressive technology calculations. Like all good architects, Pioneers try to pre-compute anything that isn’t likely to change on the fly and store it so that results can be retrieved and acted upon quickly. Believe it or not, mainframes still have their place. But the Pioneers’ tendency is to distribute computationally-intense loads across parallel computing clusters and/or to distribute the load across multiple systems and/or to perform the processing as close to the end-consumer as possible (in a rich client-side application).

Data Structures. Pioneers model objects, yet store data in hierarchical data bases (e.g., mainframe IMS), in object databases (e.g., Gemstone), in relational databases (e.g., Oracle and, more recently, MySQL). When possible, they also use large flat files (e.g., XML tagged files) to store data records for fast search and retrieval. Data structures that are of interest to Pioneers today include HP’s NeoView, GreenPlum’s massively parallel shared-nothing database, or Hadoop. In five years, distributed file systems will be de rigueur. Today, most of the applications they’re using don’t require processing of massive amounts of data. Instead, they are typically read-intensive applications that let end-users access and manipulate information locally.

Rich Internet Applications. Today’s Rich Internet Application platforms are of interest to Pioneers. They are trying out Ajax, Flex, Adobe AIR runtime, Silverlight, and/or Google WebTop (GWT) applications. Right now, based on our small sample, Google seems to have the most appeal and success.

Virtual Computing/Cloud Computing. Pioneers have been early adopters of virtualization. They like the idea of using virtual machines, both in their data centers and in the “cloud.” Virtual computing everywhere fits the Pioneers’ view of how the world works. You want to have a lot of virtual machines that are optimized for doing different kinds of things. These will probably become more and more specialized and you’ll have many of them optimized for different services. “I hand you a workload and you run it.”

It doesn’t matter where Virtual Machines are as long as they’re redundant, manageable, and secure. If one fails, there’s another ready to take its place. You can monitor and manage, observe and optimize a lot of distributed computing in the cloud and/or in your car, on your PC, or in your watch. Pioneers predict that the use of virtual machines will explode over the next few years. It’s simply the logical next step in the continuing evolution up the ladder of abstraction.

However, one of the things that Pioneers always bear in mind is that no matter how well you abstract the virtual from the physical world and/or model the events in the real world, there’s always an unbreakable connection. If you begin operating as if the virtual world IS the physical world, you’re in trouble. Physics matter. You can’t escape the fact that even in the world of virtual computing, you need to be able to start up and shut down physical as well as virtual machines.

Retail Commodity Computing Sets the Standard. You might think that these Pioneers would favor exotic systems. But it’s actually just the opposite. They find commodity PCs, Macs, servers, discs, phones, and gaming devices both powerful enough and open enough to use for many applications. They favor the use of low-cost consumer-oriented Web services and applications. Don’t spend money on expensive and exotic gear when your child’s PlayStation will do the job just fine! In fact, cloud computing fits this commodity retail model. With cloud computing you get access to networking, computing and storage available on a pay per use basis. “It moves at the speed of light, has an API, and takes a credit card.”

Biggest Constraint? One of the biggest issues on Pioneers’ radar this year is energy consumption. Computing uses energy. As you add more computing power, your energy consumption increases. Even with green data centers and more energy efficient computing and power management, they still predict that we’re going to run out of the energy needed to power the millions of machines (virtual and physical) that we use to understand and live in the world. They predict that cloud computing suppliers will lead the charge in moving to cheap energy. This will happen at the same time that it becomes easier and easier to spawn and manage many computational threads in parallel (using lots of virtual servers). “Right now, we have a phenomenal number of watts running a small number of threads.”

Make Complexity Manageable. Pioneers spend a lot of time trying to make it easier for people to do what people are good at. They think about information architecture: what are the objects and entities that people care about and what do we need to know about them? They provide simple yet accurate models of what the business is doing.

They scope problems into bite-size chunks so that human beings can wrap their minds around them and pay attention to the important things, without being blind-sided by the interdependencies of one problem space with others. The Pioneers’ habit of always thinking in terms of well-defined interfaces between interacting entities helps them keep these boundaries logical.

They design applications that let human beings focus on managing exceptions while passively monitoring what’s going on in their domain space. They describe it this way: “We remove the human middleware. We take the tedious, repetitive work away from people. We focus on awareness, rather than on doing things. We take work away and build a passive awareness of what’s happening. We simplify things for people, but don’t try to compete with the human mind.”

Are You a Pioneer Too? Does this sound like the way you view the world? If so, and if you’d value the opportunity to commune with other like-minded souls, please contact me ([email protected]) to find out more about my Pioneers group.

Comments

[Patty Seybold]

OOPS! Thanks for catching that, Patrice!
All fixed....
Patty

[Patrice]

Just a little correction, it's Silverlight, not Silverstream.

Anyway, it's an excellent post!

[Zenobia Godschalk]

Hi Patricia,

Enjoyed your last two posts and thought they were very timely, and actually quite inter-related (E-merchandising and Golden Age of IT).

As you mentioned, massively parallel databases are definitely gaining traction in the market, as companies realize what a goldmine of information they can tap into. And these databases are of particular interest to e-merchandisers, as they often only have seconds or minutes to influence a consumers’ behavior – or risk losing an additional sale. Here’s an interesting article from Tasso Argyros, CTO of Aster Data, on how e-tailers can leverage these types of ‘frontline’ data warehouses to make sure that they can take on ‘Black Friday’ and ‘Cyber Monday’ with ease.
http://www.esj.com/news/article.aspx?EditorialsID=3409&pg=1

Tasso talks about how an “always-on” frontline data warehouse enables retailers to keep their recommendation engines running, even during times of peak load. While many sys admins wring their hands this holiday season, the ‘Pioneers’ you talk about will be sleeping well, knowing their systems can handle (and even scale with) the load coming at them, without missing a beat.

Disclosure: Aster Data is a vendor who provides an always-on, always-parallel MPP database architecture with the In-Database MapReduce programming framework for business-critical applications.