Draft a Good Team
ERP systems are complex. We're talking about complicated networks performing complicated applications on complicated business processes. How do you make sure you have the best possible team to implement the system?
Blending is important. Include IT disciplines such as programmers, administrators and configuration managers with accounts-payable clerks, human-resources staffing specialists and purchasing agents. It's not as important to select a person with a particular IT skill as it is to choose someone who can make a valuable contribution to the team's direction and goals. At the same time, he or she has to be savvy in network operations. The more indispensable that person is to the operation of an application, the more you need him or her on the team.
You also need some high-muck-a-mucks on your team. It makes sense for a high-level HR manager to be included. He or she has the most to lose from an implementation that goes bad. There should also be a technical/operations lead within the team. This lead is separate from the team leader and handles the "heavy metal" aspect of computer systems and networks. A good rule of thumb when it comes to team building is 60/40: Your team should comprise 60 percent functional representation, 40 percent IT representation.
Implementation is a full-time occupation when it gets going. Once an IT person moves to the ERP team, you lose him or her from your daily operations. This is an investment that an IT manager has to make in the implementation. Implementation should not be a collateral duty of an IT employee.
Test, Test and Test Some More
To ensure that the installation works and will continue to work, you must test the system. Testing typically involves going through the delivered functionality with the enterprise's data as well as any modifications that may have been made as part of the implementation. This is generally done as a combination of manual testing (user sign-off) and automated scripting for regression testing.
The testing process normally begins with the development of a test management strategy, followed by actual load testing and functional testing. The only way to test an ERP system properly is to gather all the pertinent requirements from load and functional testing and build test cases around the user requirements gathered before design and deployment. Emulate user behavior as much as possible during the testing process. Automated load testing from the end-user viewpoint is critical.

Monitoring must be constant, especially during the critical "go live plus one month" period. The best way to monitor a system after deployment is by reusing the most critical testing scripts created during testing. You can use these automated scripts as active agents or synthetic users that run periodically to ensure that key processes continue to work.
The network should include a network-monitoring product or at least be extended to cover all elements of the ERP system. Monitoring will specify network traffic pattern trends from the ERP system and give you detailed utilization statistics. With this baseline analytic information, configuration changes can be made to tune and to optimize and get the most efficiency out of your new infrastructure.
The monitoring process will also give you insight into planning for changes in network capacity. Good monitoring will tell you where and when to add or subtract bandwidth and routing capability. It also will give you an indication of the service level. This way, you'll know about problems before users start to complain.
For example, Mercury Interactive Corp. partners with several major ERP vendors, including Oracle, PeopleSoft and SAP, to conduct testing and monitoring for enterprise clients. Mercury worked with Gates Rubber Co. over a two year period as Gates migrated its mainframe-based ERP system to a group of customized Oracle applications and third-party software packages.
|
Online Special
Worried about whether your ERP vendor will be here for the long haul?
For information on company viability and technology support, check out our online charts:
|
As a result of the testing effort, Gates was able to identify and correct a range of performance issues in the new ERP system before deployment. Performance problems in both batch and real-time situations, as well as for functional defects, came to light only under load conditions. By catching these problems before implementation, Gates saved time and effort and avoided hurting its customers.
The only given with ERP implementations is that there are sure to be problems. But problems can be minimized. Be sure to do an infrastructure audit before you begin. And make sure the team you build to implement the system has the right mix of functional and IT representatives. See to training issues, find the right consultants and make sure you manage the whole ERP process carefully. Take care of all this and a hundred other unpredictable things and your ERP implementation is sure to be a success.
Jim Romeo is a freelance writer based in Chesapeake, Va. He is the author of Net Know-How (Aegis Publishing, 2001; www.aegisbooks.com). He thanks Larry Fowler of RCGIT, Dave Crowley of Acrewood Consulting, Jack Wagner of IBM Enterprise Applications, Chris Butterfield of OuterBay Technologies and Susan Deeney of Global Knowledge for their contributions to this article. Send your comments on this article to Jim at uscg@earthlink.net.