Numerous posts and announcements have been made available in past 48 hours[1]. I don’t want to embarrass myself trying to describe what they’re about. Here I want to measure the performance penalty for the kernel patch. Kernel update has been made available on CentOS 7 (3.10.0-693.11.6.el7), roughly 12 hours behind Redhat. Here is a quick test of mysql sysbench on a small instance on aliyun, AWS, and Azure. No impact I can see on Aliyun ecs, but there is around 10% impact on AWS ec2 and Azure vm.
Aliyun ecs xn4.small before patchingSQL statistics: |
Aliyun ecs xn4.small after patchingSQL statistics: |
AWS ec2 t2.micro before patchingSQL statistics: |
AWS ec2 t2.micro after patchingSQL statistics: |
Azure vm A2v2 before patchingSQL statistics: |
Azure vm A2v2 after patchingSQL statistics: |
On a relevant note, I’m seeing the smallest instances from ali and aws beat Azure’s A2. Azure is inferior in just about everything.
The sysbench tests
sysbench --db-driver=mysql --mysql-user=root --mysql-password=xxx --mysql-db=sysbench \ --range_size=100 --table_size=10000 --tables=2 --threads=1 --events=0 --time=60 \ --rand-type=uniform /usr/share/sysbench/oltp_read_only.lua prepare sysbench --db-driver=mysql --mysql-user=root --mysql-password=xxx --mysql-db=sysbench \ --range_size=100 --table_size=10000 --tables=2 --threads=1 --events=0 --time=60 \ --rand-type=uniform /usr/share/sysbench/oltp_read_only.lua run
[1] More about meltdown and spectre
Ubuntu https://wiki.ubuntu.com/SecurityTeam/KnowledgeBase/SpectreAndMeltdown
Redhat https://access.redhat.com/security/vulnerabilities/speculativeexecution
FreeBSD https://www.freebsd.org/news/newsflash.html
AWS https://aws.amazon.com/security/security-bulletins/AWS-2018-013/
Google https://googleprojectzero.blogspot.hk/
Graz University of Technology https://meltdownattack.com/