It was designed to replace a growing set of systems that were all tiny forks of the same basic code base over a couple of years. Managing all that had become a mess, and it did need to be replaced. But there wan an opportunity and we ended up trying to reach for the stars when we should have aimed a bit lower.

The irony is that after the system had been in production for a few months, we noticed it had terrible performance that was getting worse with load.

It turned out the programmer who had written some parts of the system had it recalculating way too much data, things that didn't apply to what was going on and that couldn't have changed. It was probably an artifact from initial development (I'll do it this way to get it going, then cleanup later...). Once we caught and fixed that, it was much much faster.

That was the only optimization I remember it needing while I was there.