Million-record zones on a 2 vCPU VPS? That's not scale, that's pathology. At that point your problem is not which software, it's why are you doing this.
My upload speed is 3Mbps. I would kill for problems that need million-record zones.
Million-record zones on a 2 vCPU VPS? That's not scale, that's pathology. At that point your problem is not which software, it's why are you doing this.
My upload speed is 3Mbps. I would kill for problems that need million-record zones.
20 years. BIND 4 to BIND 9 to Knot to considering CoreDNS for k8s. The cycle repeats. Every five years someone runs this benchmark and we have this thread again. The answer never changes: test your actual workload, monitor what you deploy, don't trust vendor benchmarks.
Also:
Truer words.
What dnsperf version? There was a bug in 2.5.0-2.5.2 with TCP mode reporting inflated numbers. UDP default is fine but worth checking.