The SP system looks similar except that there are two control workstations connected to the SP Ethernet and TTY network. The frame supervisor TTY network is modified to add a standby link. The second control workstation is the backup. Figure 19 shows a logical view of a high availability control workstation. The figure shows disk mirroring, an important part of high availability planning.
Figure 19. High Availability Control Workstation with disk mirroring
If the primary control workstation fails, there is a disruptive failover that switches the external disk storage, performs IP and hardware address takeover, restarts the control workstation applications, remounts file systems, resumes hardware monitoring, and lets clients reconnect to obtain services or to update control workstation data. This means that there is only one active control workstation at any time.
The primary and backup control workstations are also connected on a private point-to-point network and a serial TTY link or target mode SCSI. The backup control workstation assumes the IP address, IP aliases, and hardware address of the primary control workstation. This lets client applications run without changes. The client application, however, must initiate reconnects when a network connection fails.
The SP data is stored in a separate volume group on the external disk storage.
The backup control workstation can run other unrelated applications if desired. However, if the application on the backup control workstation takes significant resource, that application might have to be stopped during failover and reintegration periods.