File Geodatabase file names / Updating a file geodatabase

[GeoffGrimm]

I'm distributing a file geodatabase to a number of ArcReader users.  At the moment, we're just sending out .zip files of the entire geodatabase.

The updates need to occur frequently, but in general the database doesn't change a lot during updates, so I'm investigating distributing just the differences.  The most straightforward way of doing this seems to be using patching software like RTPatch from Pocket Soft.

Due to the way we create the data, the filename of the file that corresponds to each feature class may change between versions.  That is, the address feature class might be named a0000002c.* in one version, and a000001ff.* in another.

Unforturnately, RTPatch operates on a file by file basis, so it sees the change as a delete of one file and an add of a new file.  The patch file turns out to be just as large as the original data.

So, I need some way of making the file names the same between versions of the data.

One workaround I've found is to copy each feature class into an empty geodatabase in alphabetical order.  It seems that the new file names are the same between versions of the data, but that's just an observation.

So, is there any way to force the feature classes in a file geodatabase to use the same filename?

Is there a better way to go about this?

Thanks

Geoff

[VinceAngelo]

The file geodatabase filesystem model was never really intended for piecemeal distribution
like this -- There are so many ways that this process could break down and corrupt the
remote geodatabases that I can't recommend it.

You might want to look into geodatabase replication or using the FileGDB API to transmit
a custom set of change messages (though detecting the changes would not be easy).

- V

[GeoffGrimm]

Hi Vince - Thanks for the reply.

The patching software will ensure that the target folder structure is identical to the original, so I'm not really worried about data corruption.  But I'd like to make sure that each table's file name is the same for each version of the data.  That allows the RTPatch to make the patch files as small as possible.

If the number of tables doesn't change, and the table names stay the same, will using ArcCatalog to copy the tables in order into an empty file geodatabase produce the same file names each time?

Is it possible to determine the file name that each table uses?

Geoff

[VinceAngelo]

It *likely* will be in the same order, it might even *always* be, but it's not required.
The issue is addition of new tables (removal too), where you'd get better performance
if you *didn't* use this alphabetical model.

I don't believe the dictionary tables are exposed (I've always gone on file size changes
to determine which file corresponds to which table).

In a replication model, only the changed *rows* would be transferred, not the files which
contain changed rows (and their indexes).  This is where the real storage savings lives.

- V

[LanceShipman]

The dictionary tables are not exposed. Is your primary data store a file geodatabase or ArcSDE. If its ArcSDE you should look at replication. (http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//002700000024000000.htm_)