August 26th, 2021

Machine translation is used partially for this article. See the Japanese version for the original article.

Introduction

This time we will learn about HA cluster configurations and HA cluster topologies.
Let's work hard together to become HA Cluster Master.

HA Cluster Configurations

There are two main types of HA cluster configurations: shared disk type and mirror disk type.

Shared disk type

The shared disk type stores data on a physically connected shared disk from the servers that configure the HA cluster, allowing the same data to be accessed after failover. If one server is using a specific area of the shared disk, it cannot be accessed by the others.

If a failover occurs to Node 2, Node 2 can connect to a shared disk to continue working with the data you've been using. Since there is no performance degradation in data writing, it is used in systems with a large amount of data writes, such as database servers.

Mirror disk type

Mirror disk type mirrors business data between disks on two servers so that the same data can be accessed after failover. Mirroring is the process of writing data to a local disk on Node 2 over the network at the same time as writing data to the local disk when a data write request occurs on Node 1.

If a failover occurs to Node 2, you can connect to the local disk on Node 2 to continue working with the data you've been using.
If you write data on one node(the active server), the data will have to be written on the other node(the standby server), which degrades write performance. However, you can build a HA cluster on using only the server's local disks without the need for a special external disk, such as a shared disk, so you can configure your system at a low price. In addition,When the standby server is placed in a remote site to configure a remote cluster as a disaster countermeasure, a shared disk cannot be used, so a mirror disk type is used.

HA Cluster Topologies

There are many different types of HA cluster topology, but here are two main types for operations.

Uni-Directional Standby Cluster System

Uni-Directional Standby Cluster System is the most typical configuration of operating a server as an active server and the other as a standby server. This configuration of operation is also known as ” Active-Standby”. When a business application failovers, the server that was the standby server continues to work as an active server.

Multi-Directional Standby Cluster System

Multi-directional standby cluster system, also known as ”Active-Active”, is a configuration of operation in which both servers are the active server and standby server for different operations. The active server which is running Application A is Node 1, and the standby server is Node 2, but from the perspective of Application B, the active server is Node 2, and the standby server is Node 1.

In Multi-directional standby cluster system with different applications, there is a point to be aware of.

If failover occurs during Multi-directional standby cluster system, multiple applications must be run on a single node. Therefore, note that the load is concentrated on one node during failover.

Conclusion

This time, we learned about HA cluster configurations and HA cluster topologies. In the next part, we will explain the function of HA cluster.