Chapter 1

Understanding clusters

What a cluster is

A cluster is two storage systems (nodes) that are connected to each other either directly or through switches. You can configure the two nodes so that they share access to a single set of disks, subnets, and tape drives, or you can configure them to have two distinct sets of storage, each owned by one of the nodes.

Note
You may also see the term active-active controller configuration; this is equivalent to cluster as it is used in this document.

Although most storage systems have external controller units, such as the F800 series and the FAS900 series, there are some storage systems, such as the FAS200 series, that have internal head units. The FAS270c, for example, is a cluster consisting of two internal controller units that reside in and share the same disk shelf.

About this chapter

This chapter discusses the following four types of clusters:

In addition, this chapter discusses how standard and mirrored clusters help reduce single-points-of-failure (SPOFs). For more information, see Cluster failover and single-point-of-failure analysis.

Cluster type comparison

The following table outlines the differences between the different types of clusters, and when you might want to use each type.

Cluster type	Data duplication?	Distance between nodes	Failover possible after loss of entire node (including storage)?	Notes
Standard (failover) cluster	No	Up to 500 meters (see note)	No	Use this configuration to provide higher availability by protecting against many hardware single-points-of-failure.
Mirrored (failover) cluster	Yes	Up to 500 meters (see note)	No	Use this configuration to add increased data protection to the benefits of a standard cluster.
Nonswitched MetroCluster	Yes	Up to 500 meters (see note)	Yes	Use this configuration to provide data and hardware duplication to protect against a local disaster (for example, a power outage to one node).
Fabric Attached MetroCluster	Yes	Up to 30 kilometers	Yes	Use this configuration to provide data and hardware duplication to protect against a larger-scale disaster.

Note
DS14mk2 FC disk shelves using ESH or ESH2 modules, or DS14mk2 AT disk shelves using AT-FCX modules operating at 2-Gb Fibre Channel speeds can attain a maximum node separation distance of only 300 meters, not 500.