i tried that for awhile. cpu load was getting way up there, lightning crashing, ssh responsiveness slowing to crawl, requiring daily reboot of node. increasing zram didnt help. searched web, found sthg suggesting mempool > 300 could br cause. put back to 300 and problem has largely gone away.
Discussion
Interesting, I though it wouldn't be an issue if one has RAM to spare but I guess there's extra CPU and network load associated with that.