>I have a table with over 900,000 records in it. I have used sys(3054,1) and the query is fully optimized. I also am indexed on deleted(). For example a very simple query like
>
>SELECT id from c3mid where bed = 3
>
>takes 5 to 7 seconds (about the same time it takes to browse the table). The query time is about the same regardless if 10 results or 1,000 results are returned. The are indexes on both ID and bed. If I take a similar table that has only 55,000 records and a similiar query, the result is subsecond.
>
>The large table hasn't be reindexed in about 2 months. It has 600 records added to it per day, rarely any are deleted. Could this be part of the problem?
>
>ENVIRONMENT
>PCS: NT4 and Win95 with tables on Netware 4 server with a tons of bandwith.
>Fox: VFP6 (SP2)

How much RAM? If the intermediate memory alocations durring the query start to cause swapping, performance hits a brick wall. I think that rushmore creates a bitarray for each indexed expression and performs boolean algebra across the bitarrays. This is just a guess based on observed behavior.