Combine readers for single line of processing before splitting out to writers in FME

Question

I have a relatively simple but repetitive FME workbench where I have approx 30 readers that go through much the same processing before being written by approx 30 writers (one-to-one), something like what's shown in this example screenshot:

I believe it is possible (I recall seeing this in a former colleague's file once years ago) to somehow combine the reader outputs into a single line of processing before splitting out to writers (or for any specific transformers before writing), something along the lines of this example:

However, I cannot find the required transformers to combine the inputs and for later splitting the outputs. I've searched under "join", "merge", "combine" but none of the transformers I've looked at so far seem to be what I need.

What transformers do I need to use for this, and how do I configure them to split the outputs to the correct writer path later?

Answer 1

You want the FeatureTypeFilter transformer. That will split the data back into its original layers.

Rather fortuitously, this came up during a recent webinar (during Random Acts of Transformation - where we pick a transformer at random and try to demo it). Check out this video: https://youtu.be/AoYJraqG-Rk?t=3313

The alternative is to set a fan-out on the writer. So have just one writer feature type and set it to fan-out by the attribute fme_feature_type. It depends if that is one writer with multiple layers (feature type fanout) or multiple writers (if the same format, try a dataset fanout).

But both techniques - and one or two more - are covered in detail in that video.