Depending on how bad the perf is, it may be worth it throwing the instance on a real big machine to take the load to buy time for the optimization work without turning many users away. Contributions should increase with the flood of users. If a big machine could mask the problems of course. Maybe they’re too bad to mask. You probably know.