Is there a WHERE-clause defined in the reader? If yes, are the referenced fields indexed?

No WHERE clause at all. It's just a simple read on all records in the table.

Have you tried compacting the mdb first?

How slow are we talking here - any resorsouce warnings in your log file?

Curious what platform you're on, and if you tried the JDBC MS Access reader as well. That one has greater performance in our tests. Wondering if there may be something special about your dataset too -- if you're willing to share it, please do send in to support@safe.com or post it here.

I'm experiencing the same problem. Reading an .mdb database of 200 MB with FME workbench is taking me close to 2 hours to read with the MSACCESS_JDBC reader. I've added the following environment variables in >tools> Edit Header

FME_JVM_MIN_HPEAP_SIZE: 1024M

And the improvement is quite minimum (~10% improvement).

In case it helps, mdb_process_test.txtI attach the log file where I am reading only a few MB of the database, and it still takes almost 2 minutes:

Thanks!!

Is the input file on a local drive or on a network share? Are you running a fairly recent version of FME?

I'm experiencing the same problem with a 173MB Access database. I've copied the Access database locally and compacted it for only a small performance gain.

Used ADO drive instead of JDBC driver an much much faster!

My apologies for not responding in time. That is good to know. One of the issues I had at the time (I later discovered) was network latency issues. At the time to remedy this, I simply copied the access database locally to perform my processing. So even with the JDBC driver, I experienced more efficient processing. I may try to run this process from the network location with the ADO driver to see if that makes a difference. Thanks!