>>>A loader? As to my knowledge a loader can be a very small exe and the actual code will/can be in a huge app-file. Suppose the small loader is on the local machine and the huge app-file is on the network, would that give more speed? I seriously doubt that.
>>>
>>>The advantage of not having the exe on a local machine is that an update is simpler. Yes I know that a batchfile could take care of updating, but it's not in my hands here.
>>
>>No, I meant a small exe which would copy the big exe (or .app, doesn't matter) from the server if newer than local (or local is missing), then would run the local. Each user pulls each version of the exe exactly once from the server. May take a hit when new version is up, but that's actually staggered (unless they all log in at the same time).
>
>From your discussion, it appears there is still doubt that having an EXE on a local machine instead of shared across the network is faster. With the shared exe, constant additional network traffic exists above and beyond the data and indexes.
>
>With a local exe, no regular network traffic will happen due to the shared exe. There will be an occasional hit for updates to the program - but if the exe is also broken into pieces, and only those pieces are sent, the benefits of no constant network traffic and that of having unshared, exclusive access to a local hard drive are combined.
>
>Hasn't everyone tried running a query on a local hard drive versus over a busy LAN? Isn't it obvious that an exclusive use local drive is generally faster than a shared remote drive?

Well, I did ask whether there's anything preventing the use of a loader, didn't I? Of course a loader is a loader is a loader. I think I wrote at least three :).

The additional benefit is that the "now everybody quit the app, we're installing a new version" isn't necessary unless there's something that requires exclusive opening of tables. Exe is updated whenever they get around to do it.

I think my record was running four concurrent versions. Few users were just slow, few were actually quite busy, and few were loading and unloading every few minutes, so we had two yesterday's and two today's versions running at the same time, and nobody noticed anything. Well, except those who found the bugs which made the updates necessary ;).