Recently, some people in the circle of friends have received the news of the server update. Under a cloud microblog, there are also many users who have feedback, and the cloud server is down. Then it was passed around in the circle. Some people worried that their website server would be interrupted and could not log in normally.
Server downtime refers to the server being unable to operate due to some reasons, causing the network to fail to work properly. For websites, server downtime has a big impact, and it can cause visitors to be unable to access the website.
Common reasons for server downtime are:
In the operating environment, the most common problem is the exhaustion of disk space.
In terms of performance, the most common cause is running bad SQL, but it is not ruled out by server bugs or wrong behavior. Poor schema and index design are the second most important aspects of performance.
The consistency aspect is usually caused by the inconsistency between the primary and backup data.
The aspect of data loss is usually caused by the wrong operation of Drop Table and is always accompanied by the lack of available backups.
The impact of a single server disruption is:
1) For trading companies, especially for trading platforms, server interruption is the interruption of cash flow;
2) For enterprises that have second-level requirements for local monitoring, oil temperature and oil pressure are out of monitoring. Once over-temperature and over-pressure conditions are not monitored, significant personal property damage may occur;
3) In the field of remote monitoring, monitoring data cannot be obtained and there is no record. Generally speaking, the service upgrade will be completed in about 15 minutes. During this period, the server will restart and the user will not be able to operate. The snapshot of the day may be invalid (the automatic data backup function will not have the record of the day). For a long time without restarting or upgrading the kernel, the driver has not been restarted, the restart may have problems with the file system check, related configuration changes, and startup failure. The longer the interruption time, the discontinuity of the data is meaningless for the analysis, and the interruption of 2-3 days is unacceptable.
4) IO performance degradation: After the migration, the underlying data needs to be added, so the IO performance will be reduced, and the snapshot and disk functions will be turned off. Once the data is added, IO performance, snapshots, and disk functions are automatically restored, typically 100GB of data takes about 4 hours.
Enterprises using separate servers are prone to instability, so choosing a cloud server that is armed to the high end is a fundamental guarantee for a stable operation of the monitoring system.
Reason for stability: cluster server VS separate server wins
TAOKE's server uses Alibaba Cloud's cluster server group, which is equipped with multiple hot machines, automatically switches, and is always online for 24 hours. If the platform of the monitoring enterprise adopts a single single-function server, the situation is not always online, and the disconnection is broken, the user cannot access, and the uninterrupted requirements cannot be met. The so-called dual-heater hot standby means that the function servers are set to two servers that are mutually backed up. The two servers can adopt different modes such as mutual standby, master-slave, and parallel. During the work process, the two servers will provide services to each other with a virtual IP address. According to different working methods, the service request is sent to one of the servers. At the same time, the server detects the working condition of another server through the heartbeat line. When one server fails, the other server makes a judgment based on the heartbeat detection situation and switches to take over the service. For the user, this process is fully automated and completed in a short time, so that it does not affect the business. Because of the shared storage device, the two servers actually use the same data.
Advantages of the TAOKE cluster server solution:
1. Wide range of powerful applicability and higher scalability
2, flexible load balancing and timely and effective error recovery
3, real-time heartbeat monitoring and fast and efficient drift IP
To be continued. . . .