Kontron AT8050 - Manual (Page 54)

38 AT8050
www.kontron.com
4. Hardware Management
4.1 Hardware Management Overview
The purpose of the hardware management system is to monitor, control, insure proper operation and provide
hot swap support of ATCA Boards. The hardware management system watches over the basic health of the
system, reports anomalies, and takes corrective action when needed. The hardware management system can
retrieve inventory information and sensor readings as well as receive event reports and failure notifications
from boards and other Intelligent FRUs. The hardware management system can also perform basic recovery
operations such as power cycle or reset of managed entities.
4.2 Hardware Management Functionality
The Front Blade Unit supports an “intelligent” hardware management system, based on the Intelligent
Platform Management Interface Specification. The hardware management system of the Front Blade Unit
provides the ability to manage the power and interconnect needs of intelligent devices, to monitor events,
and to log events to a central repository.
4.2.1 IPMC specific features
4.2.1.1 IPMC - ShMC interface
The principal management-oriented link within a Shelf is a two-way redundant implementation of the
Intelligent Platform Management Bus (IPMB). IPMB is based on the inter-integrated circuit (I2C) bus and is
part of the IPMI architecture. In AdvancedTCA Shelves, the main IPMB is called IPMB-0. Each entity attached
to IPMB-0 does so through an IPM Controller, the distributed management controller of the IPMI
architecture. Shelf Managers attach to IPMB-0 through a variant IPM Controller called the Shelf Management
Controller (ShMC). AdvancedTCA IPM Controllers, besides supporting dual redundant IPMBs, also have
responsibility for detecting and recovering from IPMB faults.
The reliability of the AdvancedTCA IPMB-0 is increased by using two IPMBs, with the two IPMBs referenced as
IPMB-A and IPMB-B. The aggregation of the two IPMBs is IPMB-0. The IPM Controllers aggregate the
information received on both IPMBs. An IPM Controller that has a message ready for transmit uses the IPMBs
in a round robin fashion. An IPM Controller tries to alternate the transmission of messages between IPMB-A
and IPMB-B.
If an IPM Controller is unable to transmit on the desired IPMB then it tries to send the message on the
alternate IPMB. By using this approach, an IPMB can become unavailable and then available without the IPM
Controller needing to take specific action.