David,
Thank you for the reply.

I think I somewhat mistakenly stated my problem here. This was the scenario running on a local area network.
1. Requery Remote View (10K returned).
2. Seek Key
3.1 Insert blank record if not found.
3.2 Update record if found.

This has no problem when only one instance is running but has significant performance degradation when at least two instances are running. Even just opening the target table from within VFP will cause the problem on a single instance.

We used SQLEXEC("DELETE FROM TABLE") to pre-cleanup the target table so the application only executes inserts to an empty remote view. This gave us similar performance to a single instance running.

We tracked the slowness when the RLOCK() was being executed on the target table when it is opened somewhere else. Is there some documentation regarding VFP's behavior on this scenario?

>Ramil,
>
>You are causing pretty intense disk thrashing with two processes deleting records from inside the file and appending records at the end. It is also causing header locking during the insert operation.
>
>If you are deleting the records and inserting new, why don't you reuse the records?
>
>
>>Given a table has 100,000 rows on it and is stored on a network server. We are using remote views to access the tables. We run a process from another server (where the Exe resides) that deletes 10,000 rows and inserts approximately the same number of rows. When there is only one instance of the application running this process; it takes around 15 minutes to complete. However, when there are two instances of the same application running the same process - each accessing a different set of 10,000 records, the performance quickly drops to over an hour to complete.
>>
>>I've heard that VFP don't do too well on the WAN; my question is: Is there a best practice document regarding VFP and network data?