I was hoping that BIND 9.10 would outperform BIND 9.9.4b1 on RHEL 6.5 but I was
surprised to see so much performance drop from BIND 9.10.
We have been able to send test traffic with 180K qps against 9.9.4b1 without
seeing query drops but with 9.10, the query drop rate was 18%. Both of the
numbers were obtained with 16 UDP listeners and 24 workers on identical
Sandybridge hardware with 24 CPU threads and 24G RAM. Default 12 UDP listeners
setting with 9.10 is even worse; hence, 16 UDP listeners are used for
apple-to-apple comparison.
We compiled BIND 9.10 and 9.9.4b1 with the same configuration options. I also
tried '--with-tuning=large' for 9.10 and that didn't really help in our test
environment.
I wonder what type of hardware ISC and test load used to test BIND 9.10. Are
there adjustment need to be made for 9.10 to have it perform on par with
9.9.4b1?
_______________________________________________
Please visit https://lists.isc.org/mailman/listinfo/bind-users to unsubscribe
from this list
bind-users mailing list
bind-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/bind-users
