Red Hat GLOBAL FILE SYSTEM 5.0 Instrukcja Użytkownika Pobierz pdf (Strona 6)

Preparing to Create a Cluster

Understanding the Cluster Configuration

The implementation of failover and data sharing clusters in Parallels systems is based on the

following statements:

• Red Hat Cluster Suite (RHCS) from the RHEL 5 Linux distribution is used as the clustering

software.

• In the used clustering scenario, the vz and parallels-server services act as clustered services.

These services are responsible for managing all virtual machines and Containers on your

Parallels server. If the Parallels server becomes inaccessible in the case of a hardware or

software failure or the vz and parallels-server services are stopped accidentally, the clustering

software will try to restart these services and all virtual machines and Containers on the problem

server, and if the latter is not possible, on another Parallels server in the cluster.

• A typical cluster configuration consists of 'X' + 'Y' servers where 'X' and 'Y' denote the number

of active and passive Parallels servers in the cluster, respectively. It means that you can run 'X'

Parallels Server Bare Metal installations and recover from 'Y' server failures. For example, if your

cluster contains 3 active nodes and 1 passive node, only one active node can fail over its

resources (the vz and parallels-server services and all virtual machines and Containers on that

node) to the passive node.

• The minimal hardware requirements include:

• 'X' + 'Y' physical servers

• 'X' data storages available on all nodes in the cluster

• 'X' + 'Y' ports on fencing devices

Cluster configuration examples: 4 active nodes and 1 passive node or 10 active nodes and 2

passive nodes.

• A single data storage is created per clustered service. In our case, each Parallels Server Bare

Metal installation is regarded as a clustered service.

1 2 3 4 5 6 7 8 9 10 11 ... 27 28

Brak uwag

Red Hat GLOBAL FILE SYSTEM 5.0 Instrukcja Użytkownika Strona 6