Home > Storage > PowerStore > Data Protection > Dell PowerStore: Clustering and High Availability > Cluster high availability
Every PowerStore appliance uses the pacemaker stack that is used for cluster resource management and making sure that the cluster has a quorum. The pacemaker is the cluster brain; it processes and reacts to cluster events such as appliances being added or removed from the cluster, or resource events that are caused by failures. During an appliance failure, management services are still serviced if there is a quorum.
A quorum is defined as N/2+1 appliances being in active communication. If there is no quorum, management operations are temporarily lost, but data continues to be serviced if available. The following figure shows an example of a two-appliance cluster that is servicing I/O to a host with Fibre Channel connectivity but has temporarily lost management access. In this example, there is no quorum between Appliance 1 and Appliance 2. Because the appliances are servicing I/O to the host with a Fibre Channel connectivity, there is no impact to the host. However, since quorum is lost, some management services are temporarily suspended until quorum is restored:
The following figure shows a three-appliance cluster where Appliance 3 crashes. In this scenario, quorum is met because two appliances are still in active communication. I/O remains accessible from Appliance 1 and Appliance 2. However, since Appliance 3 failed, I/O is inaccessible for that appliance.
The cluster IP address is a required field that you can set during the initial configuration of the cluster. This cluster is a highly available IP address that is used to access PowerStore Manager. The cluster IP address is typically on the primary node of the primary appliance. In the rare event that the primary appliance experiences a dual-node failure, the cluster IP address fails over to another appliance in the cluster if there is quorum. When this cluster IP address fails over, it might take several minutes to regain access to the PowerStore Manager.
The Global Storage IP (GSIP) address is an optional field that can be set during or after the initial configuration of the cluster. This IP address is a global, floating storage-discovery IP. iSCSI or NVMe/TCP hosts only require one GSIP and can discover all the storage paths for all the appliances in the cluster. Otherwise, iSCSI or NVMe/TCP hosts require a list of storage IPs so that if one IP is down, the host can try the next IP.