Glossary

The terminology used in this glossary conforms to PICMG and Telco industry terminology.


Glossary Listing

Advanced System Monitoring
(ASM)

The provision of hardware status information to the user or application program to enable an orderly shut down to be made before a hardware failure causes any damage.

Alternate Pathing (AP)

A software-driven facility that employs both redundant hardware and redundant software driver paths between a server and a disk subsystem or a network. If one path fails, AP can ensure that the disk subsystem or network is still available through the alternate path. For example, the alternate path can be a second port on an interface board, or an entirely separate interface board. See also Dynamic Reconfiguration.

availability

The ratio of the total time that a functional unit can be used to the total time that the unit is required for use.

Baseboard Management Controller
(BMC)

Manages chassis environmental, configuration and service functions and receives event data from other parts of the system. It can receive data through sensor interfaces, and interprets these data by using the sensor data repository (SDR) to which it provides an interface. The BMC maintains and provides an interface to the system event log (SEL). The BMC allows both the SDR and the SEL to be accessed from the system or from the intelligent platform management bus (IPMB). A typical function of the BMC is to measure processor temperature, power supply values, and cooling fan status. It can take some autonomous actions to preserve system integrity. For example, it might switch on a fan at a particular temperature threshold. An application interface may be provided to enable custom user-management applications to be built.
The BMC describes an abstract function, or role. It carries no definition of how the role might be implemented.

checkpoint

(1) A point at which information about the status of a job and the system can be recorded so that the job can later be restarted from that point. (2) A sequence of instructions in a computer program for recording the status of execution for restarting.

CompactPCI
(CPCI)

An adaptation of the PCI bus architecture defined in the Peripheral Component Interconnect Specification 2.1 (or later) to an electrically-compatible robust industrial form. This form specifies an Eurocard-style circuit board assembly that uses "hard metric" connectors to connect it to the enclosure backplane. CompactPCI is an open specification supported by the PCI Industrial Computers Manufacturers' Group (PICMG).

CompactPCI bridge

The PCI bridge between the system host processor and the CompactPCI bus. The CompactPCI bridge must reside in the system slot to provide CompactPCI clocking and arbitration that are only available from that slot. CompactPCI Bridges must be controllable by the system management controller to turn off clocks and arbitration.

Device Reconfiguration
(DR)

A process that is used in the CP2000 system to configure (add) or deconfigure (remove) device tree allocations and load or unload software driver modules while the system is running. It is analogous to Dynamic Reconfiguration that is used on some Sun high-end server systems with the important differences: it is not used to reconfigure memory or CPU resources and it can be used automatically in the full hot-swap and HA hot-swap cases when the hot-swap framework software is prompted by the System Management Controller. CP2000 HA device reconfiguration can also be invoked manually from a console.

domain

That part of a computer network in which the data processing resources are under common control. See PCI Domain.

drop-in

A drop-in is a code or data module that can be called by the OpenBoot PROM during system startup. It is placed in unused memory space between OpenBoot PROM and OpenBoot PROM. Most user-created drop-ins are used to initialize custom user hardware. They do not require that the user possesses OpenBoot PROM source code; only the binary OpenBoot PROM image need be licensed. Drop-ins are used to add firmware drivers for user hardware.

Dynamic Reconfiguration
(DR)

A software package that enables the administrator to (1) view a system configuration; (2) suspend or restart operations involving a port, storage device, or board; and (3) reconfigure the system (detach or attach hot-swappable devices such as disk drives or interface boards) without the need to power down the system. When DR is used with Alternate Pathing or Solstice DiskSuite software (and redundant hardware), the server can continue to communicate with disk drives and networks without interruption while a service provider replaces an existing device or installs a new device. DR supports replacement of a CPU/memory board, provided the memory on the board is not interleaved with memory on other boards in the system. Note that DR is used with Sun high-end server systems. See Device Reconfiguration for the analogous process that is applied to CP2000-based systems.

failover

The transfer of function from a failed component subsystem to an alternate one while preserving the operational state of the overall system. The functions transferred may include those of control and management.

firmware

An ordered set of instructions and data that is stored in a way that is functionally independent of main storage, for example, microprograms stored in a read-only memory (ROM). The term firmware describes microcode in ROM. At the time they are coded, microinstructions are software. When they are put into ROM they become part of the hardware (microcode) or a combination of hardware and software (microprograms). Usually, microcode is permanent and cannot be modified by the user.

Field Replaceable Unit (FRU)

A part or subsystem that may be replaced in the field or at a customer-site. Parts that are not FRUs are only factory replaceable.

Gigabit Media-independent Interface
(GMII)

An Ethernet network specification that defines a standard 1000-megabit interface between the MAC layer and either of the physical layers: 1000BASE-X (fiber-channel family) or 1000BASE-T (UTP). The GMII accommodates these physical layers without having to modify the upper layers (that is, the protocol stacks) for the particular transmission medium. The GMII is defined in IEEE Std. 802.3z-1998, which is included in IEEE Std. 802.3-1998.

handover

Synonymous with switchover. The transfer of function from a component subsystem to an alternate one while preserving the operational state of the overall system. The functions transferred may include those of control and management. Handover occurs when there is no failure in the system to prompt the transfer (compare failover).

heartbeat

A repetitive signal passed from one system to another to communicate the state of integrity or "health" of the sending system.

High Availability (HA)

The property of a system associated with a high in-service to out-of-service time ratio. This property can be engineered by reconfiguring the system "on the fly" to isolate failed elements so they can be replaced without affecting the operational condition.

host computer

(1) A computer that usually performs network control functions and provides end-users with services such as computation and database access.

(2) The primary or controlling computer in a multicomputer installation.

hot-plug

A slot must be powered down and isolated from the bus before an adapter card can be inserted. The hot-plug specification requires that board power be controlled and that means be provided to set or maintain the board in a quiescent state prior to its insertion or removal.

The method of putting the board in a quiescent state or of controlling power application to it is not defined in hot-plug but is left to the system manufacturer. The hot plug interface is defined by the PCISIG--see the hot-plug specification in the Bibliography.

hot-swap

The ability of a system element to be removed or replaced while the system hardware is nominally operating under power. This ability is usually invoked after a failure and is implemented by a sequence that steers the functions of the element to other parts of the system.

Hot-swap, as defined by PICMG, can be classified as basic, full, or HA. Basic hot-swap requires manual software sequencing to bring a card out of commission. Full hot-swap uses hardware enumeration signals to indicate board status. Software automatically decommissions the card. HA hot-swap provides for a fully automated decision tree and use of software and a System Management Controller/Hot-Swap Controller to decommission or commission a card.

hot-swap controller

The controller that takes care of the low-level sequencing associated with hot- swap.

Inter-Integrated Circuit Bus (I2C)

A serial bus developed by Philips for inter-package communications and typically used by them in TV sets. In Sun CompactPCI systems, it is used to link card elements in a system for management communications.

Inter-Chassis Management Bus (ICMB)

An IPMI/I2C bus (analogous to the IPMB) used to accomplish chassis-to-chassis management.

Inter-Host Bus (IHB)

An IPMI/I2C bus (the IPMB) used for direct communications between controllers on host boards.

Input/output (I/O)

Applies to system peripheral signals.

Intelligent platform management bus (IPMB)

A bus that carries serial communication signals that comply with the IPMI; it is used to communicate between CompactPCI circuit boards in a chassis.

Intelligent Platform Management Interface (IPMI)

A protocol interface with a protocol stack that includes link, transport, and session layers to provide reliability. IPMI resides on an I2C physical layer.

Keyboard Control Style interface (KCS interface)

One of the BMC to System Management Software (SMS) interfaces as defined in the IPMI specification (see the Bibliography).

LVD SCSI

A version of the SCSI bus that uses LVTTL (3.3 V) differential logic technology. This bus is currently specified with an 80 MHz maximum transfer rate and a maximum cable length of 18 ft.

Medium Access Control (MAC Address)

Synonymous with Ethernet address. The MAC address is a 48-bit address used to direct data-link layer transactions.

Media-Independent Interface (MII):

A specification that defines a standard interface between the MAC layer and any of the three physical layers: 100BASE-TX, 100BASE-T4, or 100BASE-FX. It can support both 10 Mbps and 100 Mbps data rates. Since the electrical signals are clearly defined, the MII may be implemented internally or externally in a network device.

nexus

A nexus driver supports a bridging connection for communication between devices on separate buses. These devices can be arranged in a hierarchal tree configuration with a number of bridges. In this case a nexus driver is associated with each bridge to handle communications with adjacent levels in the hierarchy.

nines

Used as a measure of system availability: three nines > 99.9% availability, four nines > 99.99%, five nines > 99.999%; six nines > 99.9999%;...

Non-maskable interrupt (NMI)

An interrupt that cannot be ignored. An NMI is issued for serious conditions that demand the processor's immediate attention.

Node

An addressable point on a network. Each node in a Sun network has a different name. A node can connect a computing system, a terminal, or various other peripheral devices to the network.

OpenBoot PROM

The OpenBoot PROM, or system PROM, contains code to run POST and a suite of user-accessible subsystem hardware tests. It has a Forth interpreter for custom user routines. Under a normal boot sequence, it provides a path to a system boot device which is accessed after POST completes. "Open Firmware" is defined by IEEE Standard 1275.

PCI domain

The functional entity that includes a host--usually with a host PCI bridge--and the peripherals that it controls. The domain does not necessarily uniquely include the PCI bus because this bus can be shared by multiple domains. For example, a second domain can comprise a second host/bridge element that controls a different set of peripherals on a shared bus. Separation and management of the domains is implemented by a controlling system mechanism that guarantees their mutual protection.

peripheral host

See satellite host

PICMG

PCI Industrial Computers Manufacturers' Group.

PCI Mezzanine Card (PMC)

A PMC card fits into a special PCI-bus connector designed to attach compact peripherals. These peripherals may decode a variety of I/O functions from this bus.

Power-On Self Test (POST)

A suite of tests run out of system firmware before any other code is loaded. The purpose of such testing is to check the integrity of the hardware before loading a software system.

Reverse Address Resolution Protocol (RARP)

The protocol broadcasts a MAC (ethernet) address and receives an IP address in response from a RARP server.

Reliability, Availability and Serviceability (RAS)

The general concepts associated with high in-service time systems and their simplicity of maintenance.

 

reliability

The ability of a functional unit to perform a required function under stated conditions for a stated period of time.

remote management

The action of managing a system or group of systems from a physically distant location. Remote management of Sun systems may be performed using the Sun Management Center application.

satellite board

See satellite host.

satellite host

Synonymous with peripheral host and satellite board. A satellite host performs independent tasks in response to commands from the system host. The satellite host has no accessible PCI space and is limited to controlling its own on-board I/O.

sensor data repository (SDR)

The database that the BMC uses to determine what sensors, FRU devices, and management controllers are in the system. This database contains an account of sensor locations, properties, and associations.

segment

The extent to which a backplane and cards combination can be extended by accounting for signal loading. In CompactPCI, a segment spans a maximum of eight card slots, beyond which some bridge elements (system bridge) are needed to provide expansion into another segment.

system event log (SEL)

The database of measured values and events that is created by the BMC based upon its sensor monitoring. This database resides in the host and is accessible by high-level applications.

serviceability

The capability of performing effective problem determination, diagnosis, and repair on a data-processing system.

shelf

A single physical computing system composed of one or more CompactPCI bus segments. Electrical limitations of the bridging interfaces may require that the segments be in close proximity.

switchover

See Handover.

system board computer

The processor board that connects to a backplane system slot.

system host

A system host accepts interrupts and owns peripherals. It executes user applications and decides the distribution of tasks within a system. In Hot-Swap systems the system host acts as a traffic router and functions to activate and deactivate peripheral cards (plug-in boards). It is not a CompactPCI requirement that the system host reside in a system slot, although this is normally the case. If the board resides in a peripheral slot that slot must be wired to receive peripheral interrupts from the backplane.

system management bus

A serial bus that carries data and control signals between System Management Controllers on peripheral boards and devices. Communications on this bus use the IPMI protocol over an I2C hardware layer.

System Management Controller (SMC)

There is a System Management Controller (SMC) on each card in the enclosure. One of these cards either assumes control by command or takes control after negotiation with the other System Management Controllers. The System Management Controller manages peripherals to improve the availability of the system. Through the IPMB, this entity receives information on IDs of, or problems with, cards in the system and can communicate that information with other cards or with a system host via another bus. The SMC can switch the PCI bridge, PCI arbitration, and PCI clocking on or off.

system slot

The card location in an enclosure that provides for CompactPCI clocking and arbitration. The CompactPCI bridge, which supplies these functions, must be in the system slot.

system-slot bridge

Provides clocks and arbitration. This device must be controllable from somewhere, including a controller. The system host need not reside on the same card but the card that performs the function of system host must be able to talk to the slot containing the system-slot bridge.

takeover

See failover.

Trivial File Transfer Protocol (TFTP)

A simplified form of File Transfer Protocol (FTP). It contains no password checking and is typically used to boot diskless workstations and other boot clients.