Redundant Hos (RH) support is a level of Compact PCI High Availability system that allows replacement of the hosts’ managing peripheral CompactPCI boards without changing the state of the boards.
Auriga engineers performed development of the RH support for Linux, including:
- failure recovery
- inter-host communication and synchronization
- switch-over strategies
The final version of RH layer, developed by Auriga, has the following features:
- it has been built on top of the PPS HSK and Intel IPMI management layer—and due to the nature of HA it is deeply integrated with the Linux init and PCI kernel facilities
- it has a host detection facility and allows switching peripheral boards from Active host to Backup in case of failure
- it provides RH API, which provides basic functions to control boards switchover behavior and fault tolerance policies
- it provides limited control on boards and hosts (like reset, power on/off, get healthy status and etc
- it is able to imitate the presence of peripheral boards on Backup host (fake copy of the boards population from the Active host) and thus it allows to start drivers on this boards simultaneously after any general failover on Active host
- it is integrated into Intel Net Structure SDK
- it is fully tested with Intel Dialogic System Release 6.0 for Linux OS software
Tools and technologies: Red Hat Linux Enterprise Edition, Advanced Server 3.0 Update 4, PCI, CompactPCI, Intelligent Platform Management Interface (IPMI).
You can download the full version case study here