Friday, January 27, 2006

3PE - why I use three estimates where one might do!

xProcess uses three-point estimating (3PE) for tasks and projects. Why?

Firstly what are the 3 points? Simply the best case, the worst case and the most likely. Just like when you ask how long the repair work will last and you get the reply - "I think it'll take 5 days but if we're lucky it could be 4. Or maybe it'll be much more difficult than I think. It might even take 10 days if there are unexpected difficulties."

In this case the person carrying out the work has given you more information than simply saying it'll probably take 5 days. We have a handle on the anticipated risk associated with this task. The risk associated with the estimate is captured in the gap between best and worst cases. A very wide range shows that he really has no idea how much effort is required for that
task!

The nice thing about giving xProcess 3PE's is that it give you back 3 potential end dates - the 50% end date (you're as likely to finish by then as to overrun), the 75% end date (just a 25% estimated risk of overrrun) and the 95% end date (overrun risk down to just 5%). It give you as a manager or custmoner more information to allow you to handle contingencies. These dates are sensitive to on-going updates to information about resource availability, other task priorities, task dependencies and so on, and if these change the forecast dates change immediately.

The one caution to make about the three-point estimating technique is that it comes up with such authoritative estimates for what we know is a risky and uncertain business. xProcess is only manipulating the estimates that you give it and there are a string of assumptions involved in both the mechanism of making the estimates and the higher mathematics that are used under the covers. My conclusion? Well if you can find a bookie to give you odds of 20 to 1 that the project won't slip its 95% end date I'd take it every time! However the main thing is not just to look at the forecast end date (the 50% date). If the 75% and 95% dates are much later start planning for overruns or marshall your resources to reduce the risks.

Wednesday, January 25, 2006

Two old friends ALF and BIRT

xProcess is an environment that covers a wide area since it is involved in all aspects of processes for projects. As a consequence it often overlaps or interacts with information that may be held in other tools or data sources. Its ability to integrate smoothly with information from elsewhere is a key to the products on-going success. There are 2 Eclipse initiatives that Ivis and the xProcess team are currently involved with, that will be bearing fruit in the product over the next few months. They are called ALF and BIRT.

ALF stands for the Application Lifecycle Framework. It's a Project is a subproject of the Eclipse Technology Project. Its purpose is to create a technology framework that will enable a diverse set of vendor tools, irrespective of architecture or platform, to exchange user data, manage business processes and collaborate in support the chosen ALM infrastructure technologies in use by development communities. After initial discussions with the ALF team (see Kevin Parker's Blog) we're expecting to be participating in the project and ensuring that xProcess is able to both generate ALF events and be called from ALF Service Flows. You can find out more about the ALF project from http://www.eclipse.org/alf/.

BIRT is an open source, Eclipse-based reporting system that produces compelling reports for both the web and printed formats. It provides core reporting features such as report layout, data access and scripting. It's integration with the next version xProcess means that it will be straightforward for customers to specific and generate their own documents and reports. BIRT's home page is at http://www.eclipse.org/birt/.

Both these projects will be represented at EclipseCon - an event where you'll also be able to meet the Ivis team of course, and see xProcess in action!