by John McDowall
|
|
| Technology Musings by John McDowall |
|
|
Friday, September 20, 2002 Information should be location Independent - Part II EnterpriseI had been thinking over this issue and a headline on Slashdot brought a lot of focus to my thoughts: Should Open Source Content Management Inter-operate?. This is an oxymoron similar to should network computers inter-operate, I would go further and drop the "Open Source" from the headline. This illustrates how badly we have screwed up the whole information model when this question even needs to be asked! We have developed the compunction to store everything away in transactional silos, I have been to a couple of conferences in the last week where the issues around enterprise integration were discussed. Everyone, myself included, got caught up in how to integrate the silos we did not ask why they were there in the first place. We are turning them into an institution by not questioning why they even exist - we are taking the attitude, that silos are natural and the way to solve for them is to develop complex integration software rather than removing the need. Nearly all data should be secure and stored reliably and not subject to corruption, but does it need full ACID level integrity - my bank account sure, purchase orders and other accounting information, but not too much else. Lets think of a model where everything is stored reliably, securely and with user level locking, more ACID level functions should be properties of the data not a fundamentally new way of storing information. One of the major innovations of XML is schema (including namespaces) that separate the description of the data from the data. This allows the data to be stored efficiently and multiple data models to be built separately from the data, if necessary rather than the current relational model where the schema is an integral part of the data design and must be done before the data can be stored. Today search engines can index almost a billion pages, with minimal human intervention, why do I need to design a schema every time I want to use a database. The biggest database of them all is the web and is was definitely not designed but finding information is fairly efficient, in a local intranet it is very efficient as all information can be indexed, whereas I think we are still trying to index the whole web. We appear to be digging ourselves into a bigger and bigger hole by making the technology more complex and assuming that silos are natural - it is time for information to be pervasive and accessible! posted by John McDowall | 10:58 PM Friday, September 06, 2002 Information should be location Independent - Part I DesktopOr the death of C:\ and its sibling /usr/homeOnce a user gets above a few hundred files on their desktop (unless they are very organized), finding files again becomes a major chore. I believe that the problem is that the physical storage of the information is needlessly exposed to the user. We do not think about our information as having a particular location but rather containing a particular set of information. Given the strangle hold on the industry by a few major players, who have no real vested interest in changing status quo we are faced with a problem. OpenOffice and similar projects attempt to unseat the reigning monopoly by emulating the features of the leading desktop office product. You do not catch up with a leader by playing by their rules - you need to change the game and the rules to successfully unseat the leader. Office products have not changed in 10 years - more bells and whistles but they are still the same painful products to use they were 10 years ago and they do not enhance the abilities of the user. What has this to do with location independent information: an opportunity to change the rules and deliver a new way of working with information that will deliver real value to users and differentiate open source office products from the stagnant monopolies of the past. OpenOffice has only a few steps to go to provide the necessary infra-structure for the change - it already saves files as XML. From XML to RDF is a small step. The two more significant steps are to save information through an engine similar to a search engine that will extract contextual information and then what is the interface to ask for information? An interesting example is Kartoo while a little immature is a good example of what a new way of looking at information could be. Several years ago there was a company called Perspecta which had a similar idea, I always thought they had something, I think I now see what... Extending this idea the engine becomes a personal research tool, it knows what you want and can be directed to external resources to find related information, either within the enterprise or outside to other public or private resources. Am I describing the Semantic Web, a new Office suite, or Mozilla with a rich set of plugins; only time will tell but I know this will come... posted by John McDowall | 10:13 PM Information should be location independentWe have coupled location too tightly with information, i.e. if saving a file requires saving it to a location not to a context. The context is how we think about the information not about some abstract location imposed by an operating system. Currently we are confusing how information is saved and its context. What is missing is the layer of abstraction that separates the physical storage from the contextual information about the file. As information has been tightly coupled to its location so user interfaces have propagated this error. Some of the problem has been lack of processing power to allow the abstraction to be implemented. This is only a partial reason as Apple showed with the Macintosh. At that time creating a graphical user interface used a substantial part of the available computing resources, but the result was such a major leap in usability people were willing to have less processing power available. One the first companies to talk about the idea of location independent information was Tibco, one of the basic concepts for the information bus was that users of the bus just requested information, they were not required to know its location just its context, similarly for putting information onto the bus. This simplicity has been lost, mainly ((IMHO) because we are assuming that exposing the file system is a fundamental building block rather than a mistake. The web and now the semantic web is a sincere attempt to transform this problem and create a new way of looking at information. The major step that needs to be taken is when the interface into information becomes similar to the concept of the information bus, i.e. I save it to the bus and request it from the not by its location but by the information context. A part of the context should be how transactional interactions should be and the level of security (also perhaps the certified identity of the creator - the value of the information is in many ways directly proportional to the level of trust and respect the consumer has for the producer.). I have started two additional notes on what and how to attack this problem on both the desktop and the enterprise - stay tuned posted by John McDowall | 9:54 PM Monday, September 02, 2002 Creating value and is EAI a dead end...Our industry is rife with stories of major technology investments gone wrong, and most of us have first hand experience with them. There is usually enough blame and recriminations to go round, but are we perhaps just expecting too much out of people rather than blaming the software, implementation etc. In the past several years there has been a push to develop software in "internet time", this was a concept I fought against (and in most cases lost...) as developing good software takes time. What the internet provided was a faster delivery mechanism and feedback system, i.e. you could deliver poor software faster to customers and hear about how bad it was in real-time - not a great model. I believe that tools are only as good as the people that use them, a great IDE does not make a poor developer a good developer. Similarly deploying complex enterprise software and then expecting the users to suddenly understand their business better and hence deliver value is almost equivalent to giving an average person a complete set of power tools and expect them to create a high quality custom home. We need to move away from mentality that clever software will fix all problems, software is a tool and for it to be really effective it needs to be used by well trained people. (Of course poorly designed software helps no one and this is not meant to be an excuse for poorly designed software.) Currently so much effort goes into the design of the software and systems and so little into the design of the human systems required to make it successful. This is not to suggest that we should be designing software with very simplistic interfaces, but rather we should match the software with the people using it and view them as a complete system to be optimized and designed to deliver business value. Deploying enterprise software and in fact any software should be considered as deploying a competitive weapon and as such the better people are trained to use it the more effective they will be. The time a team takes to become facile in any new tool is dependant on the amount of change from existing approaches, the learning capacity of the team, and the ability of the team to wok together. The bigger the distance to be crossed by the team the greater the possibility of failure or perhaps even worse; ending up at a different destination. This is what I think makes traditional EAI projects such high risk and why EAI is doomed to fail as a technology category. Enterprises will integrate systems and use this as a competitive weapon but only if they approach it a step at a time. To succeed requires tremendous focus and commitment to the team that is going to use the technology to achieve a new set of business goals. What is lacking from traditional EAI approaches is the focus on the business team and an incremental approach to developing success, both the software vendors and the system integrators are not orientated to helping companies achieve incremental successes as they are looking for large projects and as such tend to over promise and under deliver. The reason for under delivering is that an organization can only absorb so much new technology, processes and information at a time. Incremental delivery allows the organization to change and adapt to the new information and processes and use these new tools to succeed. Until we start putting organizational scalability before systems scalability we are not going to deliver on the potential of enterprise software. posted by John McDowall | 4:22 PM Wednesday, August 28, 2002 Are we done, is this all....Looking around at our industry I see very little new innovation, (maybe this is a sign of advancing years ;-)). This to me is not a sign of that we have invented everything and we should just go into maintenance mode and put our collective feet on the desk. It is not akin to the head of the US Patent Office who at the end of the 19th Century proposed closing the office as everything worthwhile had already been invented. I think this is more of a pause where we collect our creative energies for the next burst of innovation. I strongly believe that this will happen, the pressures are building and the groundwork is being laid. The pressures I see are that the information revolution has become an accepted fact but it is being held back by the innate complexity of todays architecture. This is both true in the consumer and enterprise space. The major innovation of the desktop is creating an open source office product, in the enterprise we are creating web services which are just an open systems approach to existing integration technologies. In both arenas it is becoming too hard to go beyond where we are today. This is the pressure for change, when there is an opportunity to dramatically change the landscape no amount of monopoly power or inertia can hold it back. I discussed this with a friend recently and he commented that Scotty in StarTrek never had to mount/search disk drives or deal with arcane information retrieval systems to get to information. So where do we go from here..... posted by John McDowall | 8:43 PM Friday, August 16, 2002 When are you a Guru ?So I just got one of those annoying headhunter calls offering me two outstanding web services "gurus". The first question that leapt to my mind was how can anyone be a guru in a technology that is less than two years old? Do I need to address engineers I work with as guru now as most have been working on web services since mid to late 2000? (The answers I got fell into two catagories, unprintable and "does that mean I get free beer?"). I actually think I like my gurus well rounded (like the Budda) and have them be deep technologists with the ability to learn new technologies quickly, like web services. So if any of you fast talking recuiters read this - do not send me web services gurus (or whatever the next next cool thing is), send me bright, motivated and articulated people want to build the future..... posted by John McDowall | 1:42 PM Thursday, August 15, 2002 From Blogs to subscription software..So I was using Radio to publish this Blog but was pretty unhappy with it for several reasons
So I have now moved to Blogger and it is way cool - simple to use and configure. Ability to log into the blog interface from any location and publish away - you would think the people at Userland forgot what the web was for. In many ways the differences between the two programs illustrate one of the problems with our whole industry - feature creep! Is it necessary to continually add features to protect and preserve a the revenue stream. I am more likely to pay for a new version of word that removes features and makes it easier to use. In fact I am moving to Linux for most of my work including desktop applications as I am really tired of paying Microsoft for new versions of Office. I am not sure if I really want to calculate how often I have paid for Office over the past 15 years, at teh same time I think I have used the same feature set. Are you a better communicator if you use transition effects in PowerPoint or format your word document in innovative ways? Subscription softwares day has arrived - I am going to pay Blogger a yearly fee to keep using the service to publish this blog and I will be very happy (until they make the interface to complex ...). I pay an annual subscription to keep my Linux machines up to date and I am very happy to do so. On the enterprise front I think SalesForce.com is going to make companies such as Siebel obsolete in the long term. This is happening because there is a well defined set of features that are required in a Sales application and if you capture them and make it easy to use and deploy why would you do anything else. So is the next wave of applications updates going to boast of fewer features and ease of use - unfortunately not but the team at OpenOffice and Xandros might think about this as a differentiator.... posted by John McDowall | 5:52 PM Wednesday, August 14, 2002
Looking at problems in a |
|
|
|
|
|||||||||
|
 |
||||||||||||
|
|||||||||||||
