Friday, September 30, 2011

Reusable Assets for Architects

Architects are fond of throwing terms around which have mixed, ambiguous or largely non-existent formal definitions. Indeed one of the great problems (still) of our profession is that people cannot agree on the meanings of many of the terms we use everyday. There is no 'common language' that all architects speak. If you want to see some examples look up terms like 'enterprise architecture' or 'cloud computing' in wikipedia then look at what's written in the 'discussion' section.

Three terms that often get misused or are used interchangeably fall under the general category of reusable assets. A reusable asset is something which has been proven to be useful, in some form or another, in one project or architectural definition and could be reused elsewhere. The Object Management Group (OMG) defines a reusable asset as one that: provides a solution to a problem for a given context. See the OMG Reusable Asset Specification. Those of you familiar with the classic Design Patterns book by the so called "Gang of Four" will recognise elements of this definition from that book. Indeed reusable assets are a generalization of design patterns. Three other reusable assets, which are of particular use to an architect, are:
  • Reference architectures
  • Application frameworks
  • Industry solutions
What do each of these mean, what's the difference and when (or how) can they be used?

A reference architecture is a template which shows, usually at a logical level, a set of components and their relationships. Reference architectures are usually created based on perceived best-practice at the time of their creation. This is both a good thing (you get the latest thinking) but can also be bad (they can become dated). Reference architectures are usually associated with a particular domain which could either be a business (e.g. IBM's Insurance Application Architecture or IAA) or industry (such as a banking reference architecture) or technology domain (e.g. cloud and SOA). Ideally reference architectures will not preordain any technology and will allow multiple vendors products to be mapped to each of the components. Sometimes vendors use reference architectures as a way of placing their tools or products into a cohesive set of products that work together.

An application framework represents the partial implementation of a specific area of a system or an application. Reference architectures may be composed of a number of application frameworks. Probably one of the best known application frameworks is Struts from the Apache open source organisation. Struts is a Java implementation of the Model-View-Controller pattern which can be 'completed' by developers for their own applications.

Finally an industry solution is a set of pre-configured (or configurable) software components designed to meet specific business requirements of a particular industry. Industry solutions are usually created and sold by software vendors and are based on their own software products. However the best solutions adhere to open standards and would allow other vendors products to be used as well. Most organisations want to avoid vendor lock-in and are unlikely to take the "whole enchilada". Industry solutions may be implementations of one or more reference architectures. For example IBM's Retail Industry Framework implements reference architectures from a number of domains (supply chain, merchandising and product management and so on).

Assets can be considered in terms of their granularity (size) and their level of articulation (implementation). Granularity is related to both the number of elements that comprise the asset and the asset’s impact on the overall architecture. Articulation is concerned with the extent to which the asset can be considered complete. Some assets are specifications only, that is to say are represented in an abstract form, such as a model or document. Other assets are considered to be complete implementations and can be instantiated as is, without modification. Such assets include components and existing applications. The diagram below places the three assets I've discussed above in terms of their granularity and articulation.

 There are of course a whole range of other reusable assets: design patterns, idioms, components, complete applications and so on. These could be classified in a similar way. The above are the ones that I think architects are most likely to find useful however.

Thursday, September 29, 2011

Plus Two More

In my previous post on five architectures that changed the world I left out a couple that didn't fit my self-imposed criteria. Here, therefore, are two more, the first of which is a bit too techie to be a part of everyone's lives but is nonetheless hugely important and the second of which has not changed the world yet but has pretty big potential to do so.

IBM System/360
Before the System/360 there was very little interchangeability between computers, even from the same manufacturers. Software had to be created for each type of computer making them very difficult to develop applications for as well as maintain. The System/360 practically invented the concept of architecture as applied to computers in that it had an architecture specification that did not make any assumptions on the implementation itself, but rather describes the interfaces and the expected behavior of an implementation. The System/360 was the first family of computers designed to cover the complete range of applications, from small to large, both commercial and scientific. The development of the System/360 cost $5 billion back in 1964, that's $34 billion of today's money and almost destroyed IBM.

Unless you are American you had probably never heard of the TV game show called Jeopardy! up until the start of 2011. Now we know that it is a show that "uses puns, subtlety and wordplay" that humans enjoy but which computers would get tied up in knots over. This, it turns out, was the challenge that David Ferrucci, the IBM scientist who led the four year quest to build Watson, had set himself to compete live against humans in the TV show.

IBM has "form" on building computers to play games! The previous one (Deep Blue) won a six-game match by two wins to one with three draws against world chess champion Garry Kasparov in 1997. Chess, it turns out, is a breeze to play compared to Jeopardy! Here's why.
  •  §Is a finite, mathematically well-defined search space.
  • Has a large but limited number of moves and states.
  • Makes everything explicit and has unambiguous mathematical rules which computers love.
§Games like Jeopardy! play on the subtleties of the human language however which is...
  • Ambiguous, contextual and implicit.
  • Grounded only in human cognition.
  • Can have a seemingly infinite number of ways to express the same meaning.
According to IBM Watson is "built on IBM's DeepQA technology for hypothesis generation, massive evidence gathering, analysis, and scoring." Phew! The point of Watson however is not its ability to play a game show but in the potential to "weaves its fabric" into the messiness of our human lives where data is not kept in nice ordered relational databases but is unstructured and seemingly unrelated but nevertheless can sometimes have new and undiscovered meaning. One obvious application is in medical diagnosis but it could also be used in a vast array of other situations from help desks through to sorting out what benefits you are entitled to. So, not world changing yet but definitely watch this space. 

Wednesday, September 28, 2011

Five Architectures That Changed The World

A favourite quote of mine from Grady Booch is:

"Software is the invisible thread and hardware is the loom on which computing weaves its fabric, a fabric that we have now draped across all of life". 

Software, although an "invisible thread" has certainly had a significant and visible impact on our world and now pervades pretty much all of our lives. Some software, and in particular some software architectures, have had a significance beyond just the everyday and have truly changed the world.

First, exactly what constitutes a world changing architecture? For me it is one that meets all of the following...
  1. It must have had an impact beyond the field of computer science or a single business area and, preferably, must have woven its way into peoples lives.
  2. Does not have to have introduced any new technology, may also have used use existing components in new and innovative ways.
  3. The architecture itself may be relatively simple, but the way it has been deployed may be what makes it "world changing".
  4. Has extended the lexicon of our language either literally (as in "I tried googling that word" or indirectly in what we do (e.g. the way we now use App stores to get our software).
  5. Has emergent properties and been extended in ways the architect(s) did not originally envisage.
Based on these criteria then here are five architectures that have really changed our lives and our world.  

World Wide Web
When Tim Berners-Lee published his innocuous sounding paper Information Management: A Proposal in 1989 I doubt he could have had any idea what an impact his "proposal" was going to have. This was the paper that introduced us to what we now call the world wide web and has quite literally changed the world forever.

Apple's iTunes
There has been much talk in cyberspace and in the media in general on the effect and impact Steve Jobs has had on the world. When Apple introduced the iPod in October 2001 although it had the usual Apple cool design makeover it was, when all was said and done, just another MP3 player. What really made the iPod take off and changed everything was iTunes. It not only turned the music industry upside down and inside out but gave us the game-changing concept of the 'App Store' as a way of consuming digital media. The impact of this is still ongoing and is driving the whole idea of cloud computing and the way we will consume software.

When Google was founded in 1999 it was just another company building a search engine. As Douglas Edwards says in his book I'm Feeling Lucky "everybody and their brother had a search engine in those days". When Sergey Brin was asked how he was going to make money (out of search) he said "Well..., we'll figure something out". Clearly 12 years later they have figured out that something and become one of the fastest growing companies ever. What Google did was not only create a better, faster, more complete search engine than anyone else but also figured out how to pay for it, and all the other Google applications through advertising. They created have created a new market and value network (in other words a disruptive technology) that has changed the way we seek out and use information.

Before WIkipedia there was a job called an Encyclopedia Salesman who walked from door to door selling knowledge packed between bound leather covers. Now, such people have been banished to the great redundancy home in the sky along with typesetters and comptometer operators.

If you do a Wikipedia on Wikipedia you get the following definition:

Wikipedia is a multilingual, web-based, free-content encyclopedia project based on an openly editable model. The name "Wikipedia" is a portmanteau of the words wiki (a technology for creating collaborative websites, from the Hawaiian word wiki, meaning "quick") and encyclopedia. Wikipedia's articles provide links to guide the user to related pages with additional information.

From an architectural point of view Wikipedia is "just another wiki" however what it has bought to the world is community participation on a massive scale and an architecture to support that collaboration (400 million unique visitors monthly more than 82,000 active contributors working on more than 19 million articles in over 270 languages). Wilipedia clearly meets all of the above crtieria (and more).

To many people Facebook is social networking. Not only has it seen off all competitors it also makes it almost impossible for new ones to join. Whilst the jury is still out on Google+ it will be difficult to see how it can ever reach the 800 million people Facebook has. Facebook is also the largest photo-storing site on the web and has developed its own photo storage system to store and serve its photographs. See this article on Facebook architecture as well as this presentation (slightly old now but interesting nonetheless)

I'd like to thank both Grady Booch and Peter Eeles for providing input to this post. Grady has been doing great work on software archeology  and knows a thing or two about software architecture. Peter is my colleague at IBM as well as co-author on The Process of Software Architecting.

Wednesday, September 14, 2011

Emergent Architectures

I was slightly alarmed to read recently in a document describing a particular adaptation of the unified process that allowing architectures to ‘emerge’ was a poor excuse to avoid hard thinking and planning and that emergent architectures, and anyone who advocates them, should be avoided.

The term 'emergent architecture' was, I believe, first coined by Gartner (see here) and applied to Enterprise Architecture. Gartner identified a number of characteristics that could be applied to emergent architectures one of which was that they are non-deterministic. Traditionally (enterprise) architects applied centralised decision-making to design outcomes. Using emergent architecture, they instead must decentralise decision-making to enable innovation.

Whilst emergent architectures certainly have their challenges it is my belief that, if well managed, they can only be a good thing and should certainly not be discouraged. Indeed I would say that emergence could be applied at a Solution Architecture level as well and is ideally suited to more agile approaches where everything is simply not known up front. The key thing with managing an emergent architecture is to capture architectural decisions as you go and ensure the architecture adapts as a result of real business needs.