Common problems and solutions for dual-machine hot standby

For enterprise-critical applications, it is important to ensure that the system delivers services consistently and reliably, resulting in high availability requirements and high availability solutions.

The following issues are common for choosing high availability solutions, especially for dual-machine hot standby scenarios:

Q: What are the high availability solutions?

A: Broadly speaking, high availability includes all the techniques to avoid interruption of system services. In general, the system high availability, often refers to the server's dual-or multiple-machine hot standby/fault tolerance.

Q: When do I need two-machine hot standby?

A: Decide whether to use dual-machine hot standby, the correct way is to analyze the importance of the system and the degree of tolerance to service interruption, so as to determine whether to use dual-machine hot standby. That is, how long your users can tolerate the recovery of services, and how much the impact will be if the service does not recover.

Q: The difference between the two-machine hot standby scheme and cluster?

A: Conceptually, dual-machine hot standby is a cluster (Cluster). Clusters generally include two categories, a pure application server cluster, each application server access to a unified database server, but they do not need to share the storage, this cluster is relatively simple, often take each server to provide services at the same time, and often at the same time using load balancing technology. Another category is a cluster of database servers (or other systems that require access to stored data such as Exchange, Notes). Clusters may include multiple servers. In addition to supporting dual-computer work, cluster software can also support multiple servers (Multi Node), deploy multiple applications at the same time, and flexibly set up takeover policies among multiple servers.

Q: Are there several ways to have high availability of database services?

A: The simplest is the way to use a standby machine. This is not really a high availability range, but it can control the amount of time to recover services after a failure in a few 10 minutes, and the standby can also be used for other purposes. This approach is a low-cost, simple solution that is more appropriate for applications that are not very important. It doesn't apply to important applications, but it doesn't matter if your application stops for a few days, or at least this level of protection.

The second one is to realize the hot standby of two machines by software. That is, do not use shared storage devices, but data is no longer a single point of storage, the local data can be to many hosts to do real-time data replication. The advantage of this approach is the savings in expensive storage device investments. The disadvantage is that either the data will be different or the speed of the database will be affected. For example, if you switch to a standby when a service is interrupted, there may be a small number of transactions that have already been completed on the host that have not been implemented on the standby. Unlike the recovery of backup data, after the standby is started, subsequent operations have been carried out, so the lost transactions are difficult to fill. Therefore, this approach applies to systems that are not very sensitive to the loss of small amounts of data.

The third is based on shared storage devices and dual-computer software to achieve dual-machine hot standby. This is the standard method of providing a second-level switch without being on duty and not losing data. Of course, the investment will be relatively high.

Q: Database dual-Machine hot standby, dual-machine is working at the same time?

A: In general small and medium-sized applications, the database of dual-machine hot standby is the main/standby mode, the main server to work another waiting, in the primary server failure with the help of dual-computer software to automatically switch to another server. And some large-scale applications for the full use of equipment, can take the way of parallel services, two servers simultaneously providing services, such as Oracle's RAC (Oracle Real Application Cluster), and support SQL Database router software icx for server parallel services