Saturday 8 December 2007

Its all about the method....

One interesting challenge for organisations implementing and integrating mixed technology and "suite" solutions is which method(s) to use. For many eBiz implementations the standard approach would be to use Oracle's Applications Implementation Method (AIM) and Conference Room Pilots. For Siebel implementations the standard approach is to use ePlan. For interfaces and integration a more "modern" approach of using community development and wikis can be employed.

Normally this is pretty simple - you look at what you are implementing and choose one approach - but one size does not fit all, and what if you have multiple work streams, implementing multiple technologies and with multiple suppliers (internal and external) delivering into the programme?

Here are some quick tips:

1. Ensure that the method(s) are agreed up front between all workstreams and suppliers and ensure that ALL parties have a common understanding of the methods, approaches and timelines. Sounds obvious but often these things are left open to "interpretation" and a supplier saying "yes I understand" does not always mean that they have the same understanding as you.

2. Get the business engaged right from the beginning. Again it sounds like project management 101 but with large, complex, multi vendor and technology implementations this is not always easy.

3. Get it on a blooming big picture so all can see how the methods and timelines join up. Visually representing the methods, approaches and milestones makes such a difference.

4. Get ready to coach and manage more than you expect. Not all suppliers are experienced in certain approaches. If you do not have lead consultants who are highly experienced in the methods you may need to coach more than you would like. Make sure you have someone who has practical (not theoretical) skills in a method and approach. Get a consultant on your side if you have done such a programme before.

5. Be strong and prepared to stay the course. It can become too easy sometimes to move from an approach because things get tricky. Joining up testing for example: System, Link, Integration, Conference Room Pilots, UAT, OAT etc can be a very complex and painful process if you have multiple workstreams and multiple technologies that may need data, test harnesses, users, environments etc. Dig deep and push on. It will be worth it. Make sure you have a good test team!

What methods work well when integrating a multi-vendor, multi-technology fabric in an organisation? Well there is no silver bullet solution. From experience I rate conference room pilots and iterative development on most technologies (including Siebel!) as it gets the user community engaged right from the start, and gets them 'owning' the solution early on - and that is invaluable, indeed a must have, for any organisation wanting to put a large enterprise solution in place.

Let me know your thoughts on this one.....

Saturday 1 December 2007

Rules, Rules, Rules

One area I have been spending a fair bit of time on recently is around rule engines.

Rules engines provide real seperation of business logic from the software coding. There are two key entities: the rules themselves and the facts that the rules work on.
The rules themselves can infer new facts from the facts that they are given which can lead to further rules being fired. This "inference engine" approach is amazingly powerful and the "chaining" of rules can lead to very powerful results with just a few simple rules.

For more about Business Rules Engines see the following article in wikipedia:

I have been developing solutions on 3 different engines recently, which I wanted to share a brief view on:

Each of these platforms essentially offer the same general problem solving approach using inference engines and the ability to seperate business logic from coding. I have been involved in the Artificial Intelligence arena for over a decade in one shape or another and these types of technologies used to cost £100k+ and be used a variety of high end applications for big business and the military. Now however the technology is essentially free (well it is through JBoss Drools at least).

I am majorly impressed with these platforms now.

I have know of Jess for many years (and indeed Oracle licenced this technology) and for those of you who still like to use LISP and reverse polish notation you will be right at home. Jess is a fantastic environment for doing RAD rule development - but you need a bit of technical skill to get anything out of it. What has dissapointed me about Jess - and think this was such a missed opportunity from Scandia labs - is it never developed a decent (or any really) UI that would take it mass market... and that is a ral shame because I really do think it is the best core inference engine out there in this space. It performance is absolutely fantastic.

JBoss Rules(Drools) on the other hand has both a performantengine and has been putting a significant amount of effort in the UI side and is coming on leaps (no pun for those in the know) and bounds in this space. From being a new kid on the block I think that this will be the one to watch closely in the coming months. It is getting more akin to the Intellicorp rules platforms of a decade ago which was a dream to use to develop Knowledge based systems (but which is sadly no longer available).

Oracle Rules is an odd one that I both love but at the same time frustrates me in that Oracle I feel has missed a trick with this which could have seem them corner this part of the market. It appears to me that Oracle are not 100% convinced in this markspace so have launched something to test the water and have put just enough effort to make a decent effort and see what the customers say. That said, what Oracle have provided should be enough for most organisations to get started. As it is built on the JESS platform it has a fantastic pedigree, however Oracle being Oracle, they have bastardised the Jess platform (I dont know why - maybe commercial reasons) and you cant access the full power of Jess but some mixed/limited set of the stack. I am about to start a proper implementation of Oracle Rules in the next few weeks so I will let you know more.

Rules Engines to me are a fantastic addition into any organisations operational platform. They can, if deployed correctly, provide increadible organisational agility and power and critically real business advantage when used within an SOA. To me these are the unsung, underutilised but future heroes of the technology fabric that all major organisations will be deploying in the near future.

Fusion and the information fabric

Firstly apologies for not doing anything with this blog for so long. I have been so busy with work and trying to setup a new business that I did not continue with the blogs.

I have been working as the solution manager for the past 9 months with a major retailer putting in a new end to end Oracle platform: Siebel for the call centre, Oracle eBiz for the finance and fulfilment, Analytics for the MI, a new J2EE website all joined up using SOA suite (ESB being key). We are doing pretty well and are moving into testing now.

One question that arises around this stack is how the heck Oracle are going to implement fusion applications in reality.

From a technical standpoint Siebel has been a dissapointment for me. Maybe this is because I have been working too long with Oracle technology, but when it comes to utilising open standards Oracle is light years ahead. It appears to me that its adoption and utilisation of XML is very poor indeed. In a SOA world this is crazy. For me Siebel has been more of a case of "form over function".

I have also been (so far!) pleasantly surprised by the SOA suite components. The ESB has been fairly robust and I look forward to seeing it in action when we move into Operational Testing and failover.

The solution we have put together is essentially Oracle Fusion - joining the suites with the SOA technology. Is it possible - absolutely. I look forward to seeing how Oracle are doing it in more detail.... but for now I am confident that it can be done... as we can see it in action today.... more on this later..