Providing Open Architecture High Availability Solutions

Providing Open Architecture High Availability Solutions
106
Indirect Detection – Detection of a fault by a method other than directly measuring or comparing
the value which is faulted. Indirect detection is used for time-based errors and where direct
measurement is difficult. For example, chassis temperature can be used to indirectly detect fan
speed or CPU temperature problems.
Indirect Notification – A notification of a fault in a component which is initiated by a second
component, typically based on a time-out or similar indirect condition. See Indirect detection.
In-line Notification – Notification of a component fault which is communicated directly to the
functional block next in line to send or receive data to/from the faulted component. This is different
than in-band communications, defined above.
Integrity – The attribute associated with a system which always returns the correct response or no
response.
Intelligent Platform Management Interface (IPMI) – A standard which specifies a protocol and
methodology for monitoring and controlling a hardware platform with or without an OS running.
IPMI - see Intelligent Platform Management Interface.
Isolation – Protecting the rest of the system from the fault by disconnecting the faulty component
(either physically or logically) and perhaps substituting a safe output value. If a safe output value is
substituted the system must be made aware of the failure, otherwise the system is not fail-safe.
Latent Fault – A fault that has not yet caused and error or been detected. A Latent fault is created
by an incorrect situation which would cause a fault under certain conditions. The fault is latent if it
has not yet been found because the conditions needed to cause it have not yet occurred. See also
Active fault.
Logical Isolation – Using software methods to isolate a component from the rest of the system.
This typically involves removing the component from I/O and process tables, but may also require
re-routing around the component by changing component addresses.
Logical system component – A group of components that acts as a single redundant component to
provide a service.
MAC – see Media access controller.
Maintainability – The attribute associated with a system which can easily be upgraded and
repaired.
Managed component – A representation of component in an HA system that can be individually
controlled and monitored.
Managed object – A representation of a component in a system that shows that component’s
attributes, methods and dependencies.
Management information base (MIB)– A data structure that holds information on how a system
is configured or functioning. These data structures are stored in memory and transferred to other
management systems as a method of communication system status.
Management Interfaces – Interfaces in a system between the components and the management
middleware.