I have a few that saved some cost. For example turning of metrics generation for _all_ host headers on NGINX instances serving millions of users.
That “feature” cost outages during security attacks and a quite significant SaaS bill, because we were also ingesting the logs of the dying NGINX instances.
Comments
I have a few that saved some cost. For example turning of metrics generation for _all_ host headers on NGINX instances serving millions of users.
That “feature” cost outages during security attacks and a quite significant SaaS bill, because we were also ingesting the logs of the dying NGINX instances.