- write a script in the file to import from the 'Old' version to the 'Upgrade' version.
- import data from a recent back-up to populkate the development file
- when ready, take the live system off-line and import to a clone of the upgrade file
- let the users go back on-line while you test the new version, off-line.
- if all is ok, repeat the process to import the live data and put it on-line.
Something that I briefly mentioned in that other thread was that you can reduce the need for importing data if you use the Data Separation method. Here's a link on how to convert your file into such a set up: Convert to Seperation Model
I would seem that Data Separation, espcially if the Data File was remote, would be less than robust. I have seen discussions on embedding versus file references, but if I understand the Data Separation concept...
PS: Sometimes putting on a seat belt can be onerous.
Referencing a reference?
Are you referring to "by reference" container data?
Don't see why that would be a problem with the data separation model.
You would reference a remote Data File and if you had a Container that used another external remote reference, i would have to be very robust on a sort.
Not a problem, but as you said in one of the links... could slow down.
Possibly, I'd want to run some tests to check that. The only data being passed from data file to interface might be just the file path text stored in the container field.
One of the interesting unanswered questions that I have about the data separation model is whether you get faster performance when both files live on the server vs putting a copy of the interface file on the hard drive of every client machine. I suspect that this will vary a lot with the particular database design and network configuration used.
In many cases, the advantages of Data Separation far out weigh the potential negatives but as with so many solutions we might come up with in database design there are pro's and cons to evaluate so it's very much a case of test carefully and thoroughly while not buying into the solution until sure that it is the best option for your particular requirements.