(Just to put this in perspective, it's probably a 10-20x increase in writes, depending on how many replicas you're using, and your read:write mix. Most people who use tracing do so probabilistically, with p < 1)
On Mon, Nov 16, 2020 at 2:09 PM Erick Ramirez <erick.rami...@datastax.com> wrote: > I wouldn't do it. It will significantly affect the performance of your > cluster so your monitoring ends up killing the very thing it's monitoring. > > Perhaps you could tell us what outcome you're after so we'd be able to > offer a better solution. Like audit logging, for example. Cheers! > >>
