NonStop NS-Series Planning Guide (H06.03+)

Planning for System Availability and Maintenance
HP Integrity NonStop NS-Series Planning Guide—529567-004
8-7
ServerNet Communication Pathway Considerations
for FRU Replacement
ServerNet Communication Pathway
Considerations for FRU Replacement
ServerNet communications within Integrity NonStop NS-series systems is point-to-point
with routers implemented in star configurations. Because of the flexibility of the system
architecture, no standard configuration or topology exists, such as the topologies for
the NonStop S-series systems. For illustrations of the ServerNet implementation in the
Integrity Nonstop NS-series systems, see ServerNet Communications Network
beginning on 3-2.
Planning for an Alternate System Disk
Consider setting up an alternate system disk to:
•
Minimize unplanned outage minutes by:
°
Having an alternate system disk configured as a backup
°
Loading the system from an alternate configuration if required
•
Avoid planned outages by configuring an alternate system disk to use in case you
must stop your main (or current) system disk.
When planning for an alternate system disk, allow enough space on the target disk for
the Kernel-Managed Swap Facility (KMSF) to operate.
For details on setting up the alternate system disk, see the NonStop NS-Series
Hardware Installation Manual.
Note. The default KMSF swap files immediately allocate 64 megabytes of swap space
and can grow to a maximum size of 128 megabytes. However, the Kernel-Managed Swap
Facility (KMSF) Manual recommends that you initially configure a swap file of 512
megabytes for each processor. Because the optimal amount of swap space varies, you
should adjust the size of the swap files depending on system use and workflow patterns.
As much as 2 gigabytes could be required for a processor with 1 gigabyte of memory.
HP recommends configuring KMSF swap files instead of using the default KMSF swap
files because many processes require more swap space than the default swap files
provide. Insufficient swap space can result in errors, process failures, and processor halts.