While bug 28413 reduced the amount of memory used by background job worker, I think that we can reduce it further yet. I'm seeing a lot of workers around 56MB RES and 104MB VIRT at the moment (although there are some smaller ones). Multiple that for multiple workers across multiple instances, and it adds up pretty quickly. Contrast that with bug 27267 where I've got an idle Zebra indexer that uses 7MB RES and 12MB VIRT. That scales up much nicer. Anyway, not a burning issue, but it would be useful to do sometime down the track. It could be an easy win for server resource usage.