Doug,

I would agree with Bob that looking for another machine/OS combination will not likely get you anywhere. The fact is, VFP is simply the fastest way possible to access local data.

I would also, with all due respect to your client, take their assertion that their code is optimized with a heavy grain of salt if they don't know that VFP doesn't run on other platforms.

I have a routine running now that takes a table of 5 million records and a table of 10 million records, and processes them through a series of conversions (SDF to table), joins, sorts, queries, and indexing. When all is said and done, upwards of 50 million records are read and written. It executes in about 8 hours on a Pentium III with only 64MB of RAM and 10,000RPM SCSI-3 drive.

When discussing the optimization of batch processing and large tables, the first thing to do is completely forget everything you know about Fox, and start over. Among the tricks are:

1. If sorting a table of 3 million records, never let Fox do it. IOW, don't do an index and copy it out, and don't do a SQL SELECT...ORDER BY. There are tricks to vastly cutting the time it takes to sort large tables.

2. The best hardware solution is to have two hard drives on two controllers. Large table copies, joins, etc. should always read from one physical drive and write to the other. Can cut real-world times by 50% or more.

3. If there are a lot of linear reads and writes, Win 98 with FAT32 is actually faster than WinNT. (For that matter Fox/DOS on OS/2 is faster than any possible windows platform, but this is not practical in most cases).

4. Avoid REPLACE ALL commands on large tables that are related into another table, much better to build a new table with SET RELATION TO and COPY FIELDS (sort of a fake JOIN).

There are many more I can share if this is a track you wish to pursue.

Best of luck...

>I have been asked to look into finding a faster computer for a VFP system. Currently, it takes almost 2 days to run a task [multiple tables, up to 3 million records in some of the tables] on a dual pentium 500 Mhz. Does anyone know how much faster this might run on a SUN 5000? [I already asked about optimizing the code and they think they have done everything they can. I don't know anything about the system -- they just have to cut the processing time.]
>
>Thanks,
>Doug