jwcolby
jwcolby at colbyconsulting.com
Wed Jun 9 16:57:02 CDT 2010
I have processes which process huge tables (65 million records are common) and update specific fields with the results of user defined functions. For example I have a function which builds a text "code" using four passed in values. These four values come from four fields of the record and the resulting code is placed back in another field of the same record. ATM I build a "cover index" just before updating this field, perform the update, then delete the index when done. I am wondering whether you guys believe that the time to build the cover index will be recovered by less time taken to perform this update. IOW does it take sufficiently less time to get data from an index vs a table scan that it is worth building the index? -- John W. Colby www.ColbyConsulting.com