Case Studies

Spirax Sarco

Spirax Sarco had several legacy systems that had been either developed in-house or had been heavily modified versions of proprietary software.  The in-house developed Sales Order Processing software was viewed as being no longer suitable and the decision was taken to purchase and implement a packaged system as a replacement.

During the summer of 1999, a brief requirements document was drawn up and a Project Manager was hired to oversee the selection process.  After viewing several suppliers, JD Edwards was chosen as the most suitable supplier and contract negotiations began.  The contracts were agreed in September 1999 and the implementation planning began.  The project was code named Phoenix.

The hardware platform chosen was HP 9000, a K360 processor, with a smaller K series as a batch server and fail-over device with Oracle RDBMS parallel server, RAID and disk mirroring to give security and redundancy.

The version of JDE One World was 7.332, client server with Citrix Metaframe.

The implementation team included a Project Sponsor, the UK Sales Director, Steering Group and Project Team.  The Project Team were drawn from the business and given over to the project on a full time basis.

Although the project appeared to be well established, and the decision was taken to change the business processes, rather than change the software, after a few months it began to flounder.  The initial date for Going Live was not achieved, as there were several hardware and set-up issues that caused major delays.  Furthermore, there were several very serious business issues that had not been resolved.  A new date was set for 2 January 2001, on the basis that they believed, erroneously, that they were not far from going live in July 2000.

The newly appointed Group IT Manager, in reviewing the project in August 2000, concluded that all was not well and advised the business that a review by an independent consultancy was a proper course of action.

CCL had already opened discussions with Spirax Sarco with regard to another, totally separate project.  The Group IT Manager asked CCL whether they could carry out a short review of the Phoenix project.

"Walter Sumner (our CCL Consultant), over the last 12 months, deserves our special thanks and congratulations for picking up the project and piloting us through some very difficult times and decisions.  Without his particular style, knowledge and determination our progress would have been much less satisfactory" commented Andrew Wilkinson, UK Sales Director.

CCL Project Case Study

The initial Programme of Work was to carry out a review of the project status and to make recommendations as to its future conduct.

It became obvious from the interviews undertaken with the Steering Group members, Project Team members and reviewing the documentation that the project was in difficulty.  The structure was good, but the participants were wrongly chosen.  The project management was weak, understandable, since the Project Manager had never been involved with such a large IT project previously.  There was no detailed project plan and the outstanding issues were business critical.  CCL drew the conclusion that the proposed go live date of 2 January 200 was unachievable.

The ensuring report and board presentation made a series of high-level recommendations.  There was insufficient data available for a possible go live date to be suggested.  CCL took the view, that a more detailed study was needed to suggest a re-structuring of the Project team members, a new Steering Group and a detailed project plan that showed each task and sub-task with elapsed time-scales and dependent events.

The Board accepted CCL's recommendations and a new Programme of Work commenced.  The result of this was a revised project structure that had a new, more senior Steering Group, chaired by a main Board member.  A revised Project Team that included Departmental Managers that represented not only the Sales elements, but he Warehousing and Manufacturing departments.  The existing team members stayed as the task group responsible for carrying out the training and system set-up as well as developing the procedures documentation and software prototyping.  The project plan was also developed to include the detailed tasks and dependent events, time phased with applied resources.

As a result of this, June 2001 was viewed as being the most likely go live date.

Shortly into the project, it became clear that the latest version of One World, Xe, was now stable and should be moved to, preferably before going live.  The decision was taken at the Steering Group meeting as this not only added to the project cost, but built in a delay.  However, it was viewed by both CCL and JDE as a wise move as it meant that no further upgrades would be needed for eighteen months after going live.

The Project implementation was made all the more difficult because the company had several legacy systems, most notably HDMS, which it wished to retain.  This meant that interfaces had to be developed between HDMS and One World.

Throughout spring and early summer, prototype scripts that reflected the business practices, were developed and the business issues that arose were identified, modelled and resolved.  During the prototyping phase, several issues revolving around the warehouse arose that became difficult to resolve.  This led to the decision to take an additional module, Transportation, in order to overcome them.  The unfortunate result was a delay in the planned go live date.

Because of holidays and reporting periods, a new date of 1 October was considered to be the optimum and the plan was revised to accommodate this new date.  In June and July a series of stress tests were conducted on the system to test the network and hardware capability in handling some 125 concurrent users with an order throughout of 300 per day with an average of 3 lines per order.  As a result of this, a further Citrix server was added as well as an additional processor and memory to the HP 9000 as it was clear that the existing configuration was not powerful enough to manage the proposed user population.

In mid-September, a meeting of the project team was convened to confirm that the planned go live date was still achievable.  The agreement as to the date was unanimous although there were concerns as to elements in the original project scope not being included in the go live phase.  The reason for not including these, Sales Engineers and Proposals, was that standard One World could not accommodate the requirements and extensive development work was required in order to achieve the business needs.  Since these areas could be part of a second phase, the business view was that the go live date should not be postponed again.

From the beginning of August, go live planning meetings were held each Friday where the fine details of the actual go live weekend was discussed.  As a result of this planning and co-operation between all departments, the actual weekend went well on the Sales and Finance elements, although the stock and warehouse areas caused some difficulty.  The stock check was not as accurate, nor as complete as it needed to be, but this did not become apparent until later in the week.  After a lot of very hard work on the part of dedicated team, Project Phoenix went live on 1 October 2001.

The inaccuracies in the stock count caused considerable difficulties, which had an adverse impact on customer service, by virtue of the expertise and hard work of the sales team, these difficulties were overcome in time.  However, it did require additional manpower to be allocated to the Finished Product Warehouse and it was some three months before real improvements in stock control and shipping were achieved.