@frederic@chaos.social
1) works slightly better, especially if you don't know a good heuristic to catch bots early on (say, checking wordpress pages or xmlrpc endpoints that don't exist, but sometimes they do exist)
2) consumes much more resources compared to fail2ban, kinda noticable on low-end VPSes
3) is more complex and can be trickier to configure for nonstandard scenarios

so, as usual, a tradeoff.