Continuing the previous part Microsoft Clustering Services (MSCS) and Exchange Co- Existence - Part 1, in this part of MSCS and Exchange Co-Existence with them we will look at some more concepts of failover clustering. Unlike the cluster components explained in part1 there are cluster objects which are visible, configurable and give administrative access to the actual cluster components. They work as an interface between cluster components and the physical objects that constitute a cluster.

Very commonly seen in cluster administrator these are the cluster object.

Nodes - These are the physical servers participating to form a cluster. All physical nodes participating to form a new cluster should be of equal hardware and software configuration.

Communication networks – Clusters depends on two types of networks, the one that they use to serve the client requests and the other that they use for their internal communications like ping requests and the quorum data transfer. Each of them is known as Public (the network that allows client connections) and Private (The network that is used for cluster internal communications)

Network interfaces – Network interfaces are the physical objects in a cluster which actually define the communication networks depending upon the IP addresses assigned to them. They work as a physical medium in a cluster to process client as well as cluster internal requests.

Cluster groups – This is one of the concepts where people mix it up with the virtual server on a cluster. Groups define the units of failover. That simply means they accommodate different types of resources under them and upon initiation of a failover these resources are moved to another physical server.

Virtual servers – In a normal standalone server scenario a client must find a physical server to connect and get the data it wants however in a clustered environment disks hosting data, network for communication to clients, etc are shared so logically a client will never find an actual physical server to connect. They will always try to find a specific IP address or the FQDN with the help of DNS queries. To make it possible clusters use Virtual servers which are not existent but they appear to existent. Virtual servers have their own computer account in active directory that appears to be a normal physical computer. They have their own IP address registered with DNS. What they differ into compared to a physical box is they change their masters from one node to another during failover and failback. Resources are sheltered by virtual servers and then ultimately by the physical node that owns the virtual server at any given time.

Resources - Resources refer to any physical or logical units on a cluster those can be brought offline and taken online, failed over to another server, can be owned by only one node at a time. They exist on a cluster in the form of a disk resource, network name resource, network card resource and so on. Cluster service manages them by communication to the cluster resource DLLs.

As an outline of all of these concepts a failover cluster can be represented as shown in the below diagram:

 
 That’s all of those basics you need to know if you have never worked on a cluster. If you have already been working on clustered servers then there is a lot to come for you as well.

In the next part of this series we will take a look at Exchange Integration and how it uses all of these concepts. You can download all of the parts of this series later when the last topic explaining how exchange uses clustering will be published. Meanwhile, we appreciate your comments, feedback and suggestions. If you have further personal feedback to be given or I can answer any questions for you, you may please feel free to write me at milindn @ msexchangegeek . com