What kind of difference in latency are we talking about?
In general, there are more moving parts, but there are a number of factors that come into play:
• base latency - is the DP in the same region as your DB? it should be, ideally
• cold starts - the proxy keeps a connection pool while hitting the DB directly from Lambdas might lead to frequent cold starts
I don’t have data right now that I can share about additional latency. I would expect some requests to be faster directly (fewer hops, warm DB connection) and some to be slower (cold start, need to open a new DB connection).
It will also depend on the DB that you are using and how expensive it is to create a new DB connection there. AFAIK, it’s more expensive on PG than MySQL.