Operational Overview

The software package consists of a combination of tightly coupled Unix drivers and utilities collectively called SavWareHA. The major components of SavWareHA are: SavWareHA, DMX Net Disk Driver, and the SavWareHA Monitor .

SavWareHA provides the administration utilities to configure and monitor all processes and events associated with the SavWareHA system.

 
No SCO Driver Modifications

Unlike other products on the market today, SavWareHA has never modified the standard SCO device drivers that operate either the master (Primary System) or the slave (Standby System) sides of the mirror. This means that any SCO supported device can be mirrored to any other SCO supported device, using any SCO supported connectivity scheme. SavWareHA imposes no restrictions on device choice other than those espoused by SCO itself.

 
Network Connection
The DMX Net Disk Driver provides the network connection for the remote half of the mirrored filesystem (s). This link is referred to as the data link , and is used to pass disk requests between the Primary System and Standby System . An additional network or serial connection may be used to provide user connectivity; this link is referred to as the IO link .

 
SCSI Bus Implementation
SavWareHA will support a SCSI link . The reason that Avnet Computer Marketing does not actively recommend this connectivity feature is the fact that any bus error on the exposed SCSI bus will (most probably) immediately cause both machines to crash. Also, if one machine must be removed from the SavWareHA setup, there is the real possibility that termination of the SCSI bus would be violated, also causing system problems.

 
Monitor (smon )
The SavWareHA Monitor , which runs as the daemon ‘smon ’, is a background process that continuously monitors both the Primary and Standby System s. All events that necessitate some change in the status of the mirrored systems will be sensed and initiated by this daemon.

SavWareHA (or more particularly, the SavWareHA Monitor ‘smon ’) may be considered to operate on the Standby System . The SavWareHA Monitor will periodically pass a request (or ‘token’) between the two machines to verify that the SavWareHA system is fully viable. This interval may be tailored by the user. Failure of the Primary System will prevent the SavWareHA Monitor (running on the Standby System) from receiving the expected response in a timely manner.

Once a failure has been detected, the SavWareHA Monitor will initiate a transition to fallback mode . This transition will involve notifying specified system administrators, performing a ‘fsck ’ on the mirrored unit(s), and enabling login sessions on the Standby System . The automatic transition to fallback mode may be set to manual mode at any time.

When the Primary System is repaired and brought back on line, recovery to a fully fault tolerant status, or normal mode , is performed through a menu option. The recovery process:

• Stops all users of the mirror (mounted on the Standby System during fallback mode )
  
  
• Unmounts the mirror from the Standby System
  
  
• Starts a regeneration process from the Standby to the Primary System
  
  
• Mounts the mirror on the Primary System
  
  
• Enables user logins on the Primary System.