I am reading and writing to an ESRI SDE database table. my workspace reads one feature class then compares it to another SQL table with possible merges from three different attributes that are tested and merged one at a time. It seems to be running fine but is just taking a long time. Not sure how or iff it can be sped up.
If you can upgrade to FME 2018, that could speed up the process.
I see a number of FeatureMerger transformers, that you can replace by the much faster new FeatureJoiner in FME 2018.
Reading the attribute data first (supplier on the first FeatureMerger) allows the setting in that transformer for "Suppliers first". That too can speed up performance. You do that by dragging that data source to the top in the Navigator window.
Hope this helps.