9.11 vs 9.16 performance OS centos7.9
Hello, I am running a cache server using bind. I am currently using bind 9.11, and I am going to update it to 9.16. However, compared to 9.11, 9.16 is only half the performance. I'm asking for a solution to anyone who has experienced the same problem as me. And I have 20 threads, but only use 1 of them. I also tried the named option -n , but it didn't give me any better performance.
Performance measurement command: resperf -d normal_20210906.txt -s 59.18.54.71 -i 1 -L 2 ?P test.gnuplot -e -r 60
bind 9.16 performance: 39100 qps bind 9.11 performance: 97333 qps
How can I solve this?
[Test environment]
- Equipment: Dell R640 (CPU Intel Xeon Gold 5215 cpu @ 2.50GHz, 10core)
- OS : Centos Linux release 7.9
- BIND version: 9.16.21
- LIBRARY : libuv-1.35.0 / openssl-1.0.2k-16/ glibc-2.17-260
- build option : starting BIND 9.16.21 (Extended Support Version) id:a8aa450 running on Linux x86_64 3.10.0-957.el7.x86_64 #1 SMP Thu Nov 8 23:39:32 UTC 2018 built with '--prefix=/usr/local' '--sysconfdir=/usr/local/etc' '--localstatedir=/var/named' '--disable-auto-validation' running as: named -u named compiled by GCC 4.8.5 20150623 (Red Hat 4.8.5-36) compiled with OpenSSL version: OpenSSL 1.0.2k-fips 26 Jan 2017 linked to OpenSSL version: OpenSSL 1.0.2k-fips 26 Jan 2017 compiled with zlib version: 1.2.7 linked to zlib version: 1.2.7
BIND 9 is maintained by Internet Systems Consortium, Inc. (ISC), a non-profit 501(c)(3) public-benefit corporation. Support and training for BIND 9 are available at https://www.isc.org/support
adjusted limit on open files from 4096 to 1048576 found 20 CPUs, using 20 worker threads using 20 UDP listeners per interface using up to 21000 sockets