p579 ~ p598.
Reason for downtime
- Of the operating environment problems, the most common problem is running out of disk space.
- Among the performance problems, the most common problem is slow SQL
- Poor Schema and index design are the second biggest issues affecting performance
- Replication problems are usually caused by inconsistent master and slave
- Data loss problems are usually caused by misoperation of DROP TABLE
Overview
Obtaining high availability by reducing downtime can be considered in two ways.
Increase uptime (MTBF)
Reduced failure recovery time (MTTR)
- Avoid single points of failure. Increase spare capacity and duplicate components. For example, load balancing, after a failure, other servers take over its load. For example, redundant network cards, routers, disks, etc. Any component that may fail at a single point must be considered . You can also use MySQL cluster, such as NDB Cluster, Percona XtraDB Cluster.
- Failover and failure recovery. Failover means that the A server hangs up and resumes the operation of A. Failover means that the A server hangs up and transfers to the B server. You can also use two methods together, A hangs up, and traffic is transferred B, restore A, then transfer back to A.