Which performs better - restricting local replication via a replication formula or readers fields.

I've been coding in and out of Lotus Notes/Domino for a few years now and have come across this situation a couple of times. Wherever I've looked on the internet, or discussed with colleagues, there seems to be an implied acceptance that if you restrict access via readers fields, then the replication is faster as the documents effectively don't exist when the replication is performed.

I recently had a project with a legacy application where we were changing the names of all the users and for various reasons couldn't just leave it up to adminp to manage the change in user names for the readers fields.

So, the decision was down to whether we reprocess all the documents and affect the readers fields (which, in this case, wasn't an easy prospect) or manage the replication via a formula. In the case of this application, the readers fields were being used solely to restrict the replication so the security provided by the fields was overkill and could be removed without issue.

I searched around the place and couldn't find anything definitive that suggested one was is better than the other, so I went about getting something more than "implied understanding".

I created a database with 100,000 records each of around 120k in size and performed local replication via both processes. In the end, the replication forumula was actually a little faster (but only minimally), so it seems that from that limited experiment that it doesn't really matter either way - in our case that meant that we could get rid of the readers fields altogether.