I'm in the analysis process of converting our 250+ GB decision support database to SQL Server 7. It will be implemented on a Proliant 7000 with 5 Fibre Channel disk arrays in a RAID 5 implementation. Two of the disk arrays will be made up of 12 9.1GB 10,000 RPM drives each, and the other 3 will be made up of 8 9.1GB 10,000 RPM drives each. There are 4 tables 20+ GB that are currently implemented as separate databases on 6.5 and the devices are on separate RAID 5 arrays, each of the tables also have non-clustered indexes. I want to implement this as one database on SQL 7, and use files and file groups to spread them across the 5 distinct arrays. I have several thoughts on how this could be done, but would like anyone elses opinion on how they might implement these structures. One thought was to create several file groups that each contained 5 files 1 each on the distinct arrays, and then created the larger tables on these file groups, and put the smaller tables on a separate file group and the non-clustered indexes on a separate file group? Another way might be to put each of the large tables on there own separate array, and the smaller tables and non-clustered indexes on different arrays? Does anyone have any thoughts(for or against) on these implementations, or any thoughts on their own. Thank you for your time