Cluster type and why clusters are used

There are two types of SQL Server Clusters: Server Load balancer and failover. In the short article in The InformIT SQL Server Reference Guide, we discuss the causes of the cluster and some system requirements:
Server Load balancer Cluster
In an actual Server Load balancer group, all servers are called nodes. A group of computers are called arbitration, the computer or software service that gives external users the impression of a single server is called the arbitration manager, which transfers the processing requirements to one or more servers. This way of sharing work creates a very powerful virtual computer. If one of the nodes leaves the cluster, the arbitration manager forwards the workload to another server.
Although this cluster method sounds good, SQL Server does not support this method. Unfortunately, it does not work. It doesn't matter. Let's continue.
Failover Cluster
Failover clusters are easy to set and provide high security for the environment. In a Failover group, two servers share the same storage device. The server establishes a single signal connection with the storage device. A single signal serves the same purpose as heartbeat, so that the server cannot hear each other. The second node takes over the identity of the first node. All these processes are complete in processing consumer data. Your users may not even be aware of this phenomenon based on the application's tolerance for connection timeout.
Both SQL Server 7 and 2000 support this type of cluster. In fact, SQL Server has two modes in this cluster: active/active and active/passive. Now let's take a look:
Active/active mode
In this form of cluster, each server works independently and can also handle faults of other servers. Assume that we have two server nodes named Service server A and server B in the cluster. The cluster uses another name, cluster 1, to calibrate itself. You can use cluster 1 or server B, if server A crashes, server B becomes cluster 1 and server B. It uses two identifiers.
An active/active cluster can have up to 16 instances connected to four nodes. We will discuss the instance topic in another article, however, you can regard the installation of SQL Server on a system as an instance. In ODBC and other connection types, the name is the name of the server connected to the instance, that is, the server name + instance name, rather than the server name you are used.
Active/passive mode
In this mode, the cluster is the only device that can be seen outside, and other devices are in the "standby" status. The cluster will start only after you manually intervene or when the first group of clusters fails.
Why cluster?
You may think that the only reason for using the cluster is for security. After all, I said that SQL Server does not support real load balancing. What else do you think? In addition to the benefits of high efficiency, you can also manually transfer the faulty node to the second node, upgrade the first node, then transfer the system back, and then upgrade the second node. This may sound less important, but if you have a system that does not allow any downtime, this function is priceless.
Requirements
No matter which failover cluster you choose, pay attention to the following points. First, you need to install the SQL Server Enterprise Edition. You can also use the programmer version, but we will discuss it in this article according to the Enterprise Edition.
You also need to install Microsoft's cluster service (MSC), which is another big topic beyond the scope of this article. The link to Microsoft's cluster home page can also help you understand. The operating system must be Windows 2000 Enterprise Edition, data center edition, or a system version of Windows 2003 Server. The Windows 2003 Server has made many changes to the system, so make sure you fully understand the operating system of the cluster you want to run.
Two NICs are required for each server, which is not required technically. However, because heartbeat is required between servers, it is best not to share resources. Second, you need to share a hard disk for service sharing. Microsoft calls this a shared SCSI bus ". Most people use the storage Region network, but even if you have not read the hard part compatibility list, you must follow the instructions. Microsoft has had an accident for a long time, so when searching for the cluster and hardware compatibility list, confirm that your hardware should be in it; otherwise, the operation will not go smoothly.
You can read more steps about cluster installation in the Microsoft support center.