Case Study: FICO Implements Tape SAN Page 2
Phase 1 of the migration project incorporated confirmation of the hardware selection, designing and implementing the hardware infrastructure, preparing migration plans, and beginning the actual data migration with the eight largest databases. Phase 2 involved migrating the remaining 220 servers, stabilizing and tuning the SAN, and completing the training and handover.
The project undertaken went as follows: revisiting existing backups and archive requirements on the mainframe, and then categorizing these requirements by business needs with the aim of simplifying and consolidating the underlying business processes. Only when that step had been fully completed did FICO feel that it could consider a move to a new architecture.
"Getting right into our business processes to clarify business unit assumptions and validate the pre-existing SLA's was fundamental," says Wiltshire. "We could then even out the current incremental and full backup practice we had ended up with over time." He reports that this approach allowed the company to better schedule and automate backups to meet business demands. Further, it permitted FICO to centralize the control of all jobs at the same time.
FICO completed Phase 1 during a three-month period and Phase 2 one month later. Wiltshire confirms that he completed the project on time and on budget, and believes it to be one of the few successful consulting projects FICO has undertaken. The main benchmark of success set at the project outset was backup reliability, and this went from 60% to 98% daily reliability, an especially impressive feat when you consider the backup volume of 25TB per week with a system capability of 1TB per hour.
"To succeed, we learned that you have to have the right expertise available, especially if you are using a new technology", says Wiltshire. "Have your team involved in the entire project and set clear expectations and guiding principles."
In addition, he considers it essential to fully verify that assumptions and requirements are based in fact and not emotion in order to control the project scope. Last but not least, his advice is to always plan for the future.
"Planning for the future was the lesson of our earlier SAN efforts in '98-'01, and served as hard-learned but invaluable experience for this project, which has been an undoubted success," concludes Wiltshire.