Windows Server Failover Clustering is mainly used for what the name is, meaning that you can create a cluster, bunch, for example, a bunch of servers that are identical and perform the same way, one like the other, to handle the same task or connection or request. And it allows you to create a virtual unit that contains multiple identical units. For example, if you want to use virtualization with Hyper-V and you want to deploy, five notes or five servers each one of them hosts the Hyper-V service for virtualization, and you want a virtual machine to run on every node of the cluster, then you will need to deploy or to use Windows Server Failover Clustering that you will create a virtual unit from, or that is based on physical units. This is for servers. And also you can create clusters of applications. For example, if you want to use a cluster of IAS services, you need to also use the Windows Server Failover Clustering. For example, to have a cluster of my MySQL database, you also need to use this service. If you want to use or deploy storage or file server servers, a cluster of file servers, you also need this service, meaning that each time you want to use high availability and failover Clustering, you need to deploy Windows Server Failover Clustering.
The solution has many good features. Some of the advantages, are it can give you the ability to choose... To just explain something. When you have a bunch of servers, for example, that are part of cluster configuration, each server will perform identically like the others. In this type of configuration, you need to elect one master that will handle directly the request. Meaning that all the other servers will not handle anything but they will stay in a listening mode or like a slave. And they will communicate each time they will do a hard bit on the master to see if it's still running or accessible. If they detect that the master is not responding, then they will elect another master that will handle the communications. So this feature is done automatically within the server, this service, meaning Failover Cluster service. And also you can choose which one you want to elect as a master. You can do that manually and you also can do it automatically and you can configure it so that you can have a preferred one. Meaning that each time there is a revolt of the servers or revolt of this cluster, they will communicate with them. And depending on the preferred one, this one will be chosen as the master instead of doing an election within the server.
The solution has some downsides related to the election mechanism, meaning that if the servers or the cluster services or cluster server elect one server, they will attribute to that server an object that is called a quorum. This quorum does the server or the service that has that quorum, it is taken as the master. So when you have, for example, something happens in your data centers like a power outage or an unexpected shutdown, and you can't start the master server what sometimes will happen, is since the shutdown was unexpected, the cluster didn't have time to do some post configurations before the extension of the shutdown of the servers, sometimes that quorum can't be removed from the crashed master and the cluster wants stuff. This is part of the architecture of this service, the Failover service. So sometimes we face this type of problem. In the past, we had a lot of power outages and when we started after, our server, we faced this problem. Meaning that the cluster won't stop because one of the servers, the master crashed, and they didn't want to start. So we need to access and do some line configurations so that we will force the cluster to elect a new master and ignore the crashed one.
I have been using the solution for seven years.
The stability is good, except for when something crashes or shuts down unexpectedly, then it can be a bit difficult to reestablish the service in a stable state. You need to do a lot of work.
The solution is scalable. If you want to create a cluster of servers, you need to install and deploy the Windows Server Failover Clustering service in each one of them. When you join one server to a cluster, it'll be a switch forward for the other server to join the cluster. You only need to configure the new servers you want to add to your cluster.
The technical support is great. It depends if the incident is critical or minor.
Since we have a contract with Microsoft, we are mainly using their products. I think it's the only product that is available on the Windows server. It's better to use it rather than another clustering service.
The initial setup is straightforward. It's easy to deploy the service, to activate the service. Once you understand how it works, it'll be easy to configure any type of service that you want to create a cluster from. Some scenarios require a lot of work in order to create a cluster with assistance from Microsoft, meaning software or installation or configuration assistant, which makes things easy. Sometimes you need some steps that are not well explained. Meaning that sometimes you need to check or uncheck some option boxes, and you need to know what will happen if you check or uncheck these options and this type of misconfiguration or lack of documentation in the configuration process. Sometimes you lose all your data. This is just a technical thing, you will get it easily. This solution was designed first to create a file server for storage, to store files. So when you want to create a cluster and you need to choose or assign the servers you want to join or create clusters from them, you choose server one, server two, and server three. These three servers will be a part of the cluster. In one step, there is a check box. If you leave it checked, it will take all the disc space available on the servers and format direct. Meaning that if there is something that is hosted on these servers, they will be wiped out. Why? Because by default it'll configure these servers like file servers. It is important to know that if you have databases on these servers, everything will be lost. And this step is not well explained in the configuration process in the assistant. If you want to create a cluster, you need to start with freshly installed servers and not existing ones. The setup process is a bit confusing and on some steps, there is not much information about the next steps, and what will happen. Sometimes, normally and logically, you don't need to execute anything unless you click on the Finish button, not the Next button. With the configuration assistance of this solution, some action will be applied when you click on Next, even if it's not the last step.
The deployment is completed in-house by either myself or my colleagues.
It's like IAF, it's a part of the Windows server license. So if you acquire the Windows server license, you will also get all of the Windows services licenses. It's not a product that you can download and install. It is the feature that you activate in the Windows server.
I give the solution an eight out of ten.
The solution is like IAS, the version depends on the version of the window server. So it's part of the server. So you don't use or specify the version except with which Windows server you are using.
The solution has some downside frankly, that it needs a lot of configuration and steps in order to create your cluster. This means that deploying this service is not that hard. It is easier than IAS, but if you want to configure the cluster service to create a cluster of an application, then it'll depend on the application you want to create a cluster with. But once it is done, it'll perform very well.
Sometimes you have to manually assign a master, especially when there is an unexpected shutdown caused by another problem. Sometimes you will face these types of problems. You need to manually assign a master. If you reboot the servers one by one, you don't need to reboot them at the same time. When you reboot them one by one, for example, if you reboot the master, before the master restart, it will communicate and handle the quorum to another one. It's like, "Okay, I'm not the master anymore. Choose another master that will handle the communications and requests." And when it starts, it will no longer be the new master. It will be another slave that will wait for the master to fail so they will also do an election to set the new master and give it to the quorum.
The deployment time depends on what you want to create on the cluster. Meaning if it's a Microsoft SQL server, it needs a lot of steps or a lot of configuration because you need to create a cluster, you need to create the listener because you have a bunch of servers. Meaning you have a lot of IP addresses that the master will need since the master can change. As an example, for server one that has an IP address, one is master, it will handle the communications, but once it fails there will be another master with another IP address. We don't each time teach the IP address to communicate with the cluster and that's why the Failover service will create a visual IP address that is called a listener so that it will communicate with the clients and other applications and it will handle the back of it. This means if something fails, the switching from one master to another, will be transparent to users. Since you are communicating with the virtual IP risk address, that will not change. Behind this listener, a lot of things can happen. Master fails, unless the new master changes IP addresses, et cetera. So it'll be transparent for the users. These configurations can take time depending on the application, the architecture, or the scenario you want to use this service with. If it's servers, it'll take some time. It depends on how many servers you want to create, and what type of cluster you want to create with the service or database, file system, servers, et cetera. So it can take time but not many days. You can do it within the day to have your clustered application or servers fully working.
The solution requires two System administrators to maintain.
I suggest if you want to use the product, start with a freshly installed server. Meaning that the server doesn't need to host any data from the start. Once it is configured you can put your data onto the server. Otherwise, you need to be careful in the cluster configuration process, because once you add your servers at one step, I don't recall, when you click on Next, it will configure the cluster as file servers. Meaning that it will take all remaining storage capacity on each server and create the pool of storage with the capacity of some of these remaining stores. There is a checkbox. I don't remember if you need to check it or uncheck it. Either way, you need to do the contrary of what you found. If it's checked, you need to uncheck it. If it's unchecked, you need to check. I don't recall at which step you will find it, but you need to be careful and you need to read about the service before starting to create one. And also you need to read each document so that you will know what are all the requirements to create a cluster.