>I want to be able to run Visual FoxPro programs against data that sits on a server in California. I live and work in Colorado.
>
>The VFP programs typically crunch lots of data. SELECTS or SCANS on 50 MB tables would not be unusual. Appropriate index tags are often in place and are currently helping with the process. At the end of these VFP programs, they may put out extract files or reports.
>
>I want to avoid copying large files. I also want to avoid running from a VFP in client in CO against large data files in CA. Both of these activities are slo-o-w and and tend to choke the WAN.
>
>I've given some thought to a variety of ways for running VFP programs remotely. These are:
>1) PCAnywhere.
>2) Citrix Server.
>3) Moving the data into SQL Server and using client/server technology.
>4) Using AT or another Windows NT scheduler to launch a VFP program to run on a remote server.
>
>I don't know all that I'd like to about these various "solutions". There may be even more ways to do it. I hoping for comments on the relative strengths and weaknesses of the different ways to go.

You've already hit on the main point, which is to avoid pumping 50MB files over the wire.

For any such scheme to work, CA needs to let you run a VFP app either directly on their server or on a LAN-attached workstation. If they're willing to do that, you have a number of options.

It partly depends on the type of analysis you're doing. If you're doing the same daily reports over and over, a scheduler would work fine; you'd just need to dial in when you know they're done, and download the results. If the programs require human TLC to run, you'll need to use a product that lets you establish a session with CA, so pcAnyWhere might be a good option.

If you want to run custom analyses, that requires the ability to submit jobs against a server/process in CA.

There are at least a couple of options you haven't mentioned so far. If session-oriented is desired, you could set up a COM server in CA (actually, DCOM) and call its methods remotely. Or, you could set up a general-purpose analysis engine in CA. In this latter case, you could submit .PRG files to a folder on CA's server. The CA engine would periodically poll this folder. If it finds a .PRG, it would compile and run it. The .PRG could specify output folders and file names. This second option would be fire-and-forget, much like the scheduler option, but it would also allow custom analyses.

As far as products go, for remote access pcAnyWhere is the cheap & cheerful method, simply requiring an available workstation in CA. Citrix is effectively a multi-pcAnyWhere box and is fairly expensive to buy & set up. If the data are not already in SQL Server, that is probably the most expensive (time AND money) approach.