Clustering Exchange Server | Resilient Email

A server cluster is a group of independent computer systems, known as nodes, working together as a single system to ensure that mission-critical applications and resources remain available to clients. This document is a brief overview of how clusters can be deployed and what they are intended to do, especially with regards to Exchange server.

There are three primary benefits to using server clusters. Server clusters are highly available, scalable, and easy to manage.

Highly available: Server clusters provide a highly available system for deploying applications. You can use server clusters to protect against failures of hardware, operating systems, device drivers, or applications. If one of the nodes in a cluster is unavailable as a result of failure or maintenance, another node immediately begins providing service (a process known as failover). Server clusters also allow you to upgrade the operating system and application software with minimal downtime.

The level of availability required by companies varies, but it is not uncommon to require 99.99% uptime, which equates to approximately one hour of unplanned downtime per year. With careful planning and design these figures are achievable using clustering technology.

Scalable: Server clusters provide scalability for important resources and applications. When the overall load of a cluster exceeds its capabilities, you can incrementally add additional nodes to the cluster. In Windows Server 2003, server clusters can include up to eight nodes.

Using an active/passive cluster design, user levels of 3000 to 5000 are possible. Mailbox size will depend on the size of the storage resource (typically a fibre based SAN). A 200Mb mailbox limit is typical.

Easy to manage: In a server cluster, you can quickly inspect the status of all cluster resources and move workloads around onto different nodes. Because you can move processing to alternate nodes, you can perform rolling upgrades on the servers. In a rolling upgrade, a server cluster continues to provide service while software is being upgraded on each node until all nodes have been upgraded.

Server consolidation is another goal required by companies. Many Exchange 5.5 servers could be consolidated into one Exchange 2000 or Exchange 2003 cluster.

Clustering is used in a Microsoft environment for file and print servers, and cluster aware applications, such as SQL server and Exchange server. Clusters directly help reduce single points of failure in a server design. Clusters can also help reduce planned downtime. For instance, if it is necessary to take a server down to apply service packs or bios upgrades, one node can be worked on while the other(s) still provide the clustered service to the users.

Each node has one or more physical disks used for storing the operating system, the swap file, non-shared applications (such as Exchange binary files), and so on that only they have access to. At the same time, every node is attached to one or more shared cluster storage devices (such as a Storage Area Network) that contain the cluster quorum drive and Exchange shared resources (such as log files, public stores and message stores). Clustering allows users and administrators to access and manage the nodes as a single system rather than as separate computers.

In an active/passive cluster configuration, only one node of the cluster is active at any given time. All resource groups reside on that node. If that computer fails or is taken offline, the other node will gracefully take over all resource groups. The problem with this configuration is that one node is always idle.

In an active/active cluster configuration, each node of the cluster will own a group (for example, an Exchange virtual server) so all nodes are working all the time. With this approach, no nodes are idle. With Exchange 2000, the active/passive tended to be the preferred configuration, as under heavy load, there could be virtual memory problems with an active/active configuration, which sometimes gave problems failing nodes over.

Clustering can be extremely valuable for larger companies, or indeed in any company where file servers or applications like Exchange are considered mission-critical. In other words there will be a severe impact to the business if the service became unavailable for any length of time. A well designed clustered solution should provide high server availability and of course, low downtime. Exchange clustering requires the Enterprise version of Exchange 2000 (or Exchange 2003) and Windows 2000/2003 Advanced or Datacenter version.

To create an Exchange 2000/2003 cluster, you create a Windows cluster group and then add specific resources to it. The logical servers created by Exchange clusters are referred to as Exchange Virtual Servers. Unlike a stand-alone (non-clustered) computer running Exchange , an Exchange Virtual Server is a cluster group that can be failed over if the server itself fails. When one computer in the cluster fails, one of the remaining nodes in the cluster takes over for the failed Exchange Virtual Server and clients can access this server using the same Exchange server name.

The most important disk in the cluster is the disk designated as the quorum disk resource. The quorum disk resource maintains configuration data in the quorum log, cluster database checkpoint, and resource checkpoints. The quorum disk resource also provides persistent physical storage across system failures. Because the cluster configuration is kept on this disk, all nodes in the cluster must be able to communicate with the node that owns it.

The figure on left shows a quorum disk. If the servers in an Exchange cluster are running Windows Advanced Server, you can create either an active/passive or active/active cluster. The figure below shows an example of a two-node cluster topology that you can use with Windows 2000 Advanced Server. Both cluster nodes are members of the same domain. The cluster nodes are connected to the public network and a private cluster network. If only one cluster node owns one Exchange Virtual Server, this is an active/passive configuration. If both nodes own one or more Exchange Virtual Servers, or if either node owns two Exchange Virtual Servers, this is an active/active configuration.

exchange cluster A well designed cluster solution can provide the reliability, scalability and manageability that companies need. The opportunity to consolidate servers can also be a great driving force. Exchange 2003 is ideally suited to clustering - a well designed solution will give great resiliance and increase service uptime.