Glossary

access control list

(ACL) A file that specifies which users can access a particular resource, such as a filesystem.

accounting

Keeping track of resource usage on a machine. SRM provides accounting features.

ACL

See [access control list]

administration Tool

A GUI tool for configuring Solaris Bandwidth Manager.

administrative domain

A collection of network elements under the same administrative control and grouped together for administrative purposes.

ADOP

See [automatic degree of parallelism]

alarm

The means by which notification is sent when an exception occurs.

attaching

See [DR Attach]

Alternate Pathing

(AP) A software mechanism which works in conjunction with Dynamic Reconfiguration (DR) to provide redundant disk and network controllers and their respective physical links. The main purpose of Alternate Pathing is to sustain continuous network and disk I/O when system boards are detached from a machine or DSD (in the case of the Starfire) that is running a live copy of the Solaris operating environment.

AP

See [Alternate Pathing]

application resource measurement

(ARM) A means of measuring the end-to-end response time of a system.

ARM

See [application resource measurement]

ASE

Sybase Adaptive Server Enterprise.

automatic degree of parallelism

(ADOP) A feature of the Oracle8i Database Resource Manager that attempts to optimize system utilization by automatically adjusting the degree of parallelism for parallel query operations.

backfilling

The execution of a job that is short enough to fit into the time slot during which the processors are reserved, allowing for more efficient use of the available resources. Short jobs are said to backfill processors reserved for large jobs.

BBSRAM

Boot Bus Static Random Access Memory.

blacklist

A file that enables you to specify components, such as system boards, that should not be configured into the system. The blacklist file is read and processed at bringup time.

BMC Best/1

BMC Software's BEST/1 products provide tools to address performance management requirements across OS/390, Parallel Sysplex, SAP R/3, Unix, Windows NT, VM and AS/400 environments.

CBQ

See [Class Based Queuing]

CCMS

A tool that provides information to SAP R/3 allowing it to measure the performance of key user transactions and the response time of the backend database for applications.

CICS

See [Customer Information Control System]

CIM

See [Common Information Model]

Class Based Queuing

(CBQ) The underlying queuing technology used in Solaris Bandwidth Manager.

Classes of Service

A feature supported by Solaris Bandwidth Manager that allows network traffic to be organized so that urgent traffic gets higher priority than less important traffic.

classifier

A component of the Solaris Bandwidth Manager that allocates packets to a class queue. When a packet arrives, the classifier analyzes the packet protocol, ToS value, URL information, source information, destination information and allocates the packet to a class queue where it waits to be processed.

CLI

Command Line Interface, as opposed to Graphical User Interface (GUI).

cluster

A collection of computers interconnected via a high-speed interface that allows the environment to behave as a single unified computing resource.

clustered cache

A method of caching web pages where multiple servers use the intercache protocol (ICP) to talk among themselves and form an explicit hierarchy of siblings and parents. If the load would overwhelm a single server or if high availability is important, multiple servers are configured as siblings. Each sibling stores data in its cache but also uses ICP to search the caches of other siblings.

CMIP

A scalable OSI-based network management protocol that is used in situations where SNMP is not powerful enough.

codepoint

A construct used by the ToS facility of the Solaris Bandwidth Manager. The way that the DSCP bits are set in a packet is called its codepoint, and network devices translate DSCP codepoints to device configurations. For example, a DS compliant router that has two priority queues (one for regular priority traffic and one for high priority traffic) could be configured to have several codepoints map to the high priority queue, and several others to the regular priority queue.

Common Information Model

(CIM) A metamodel based on the Unified Modeling Language (UML) that supplies a set of classes with properties and associations. The Common Information Model provides a conceptual framework within which it is possible to organize information about a managed environment.

control interval

In control theory, the rate at which measurements are made and corrections are applied.

CoS

See [Classes of Service]

Cross-System Coupling Facility

A WLM component that communicates policies, metrics, and control data between Sysplex nodes.

Customer Information Control System

(CICS) An interactive transaction processing system from IBM.

DDI_ATTACH

A function, used by DR (called from the dr_driver) that provides the ability to attach a particular instance of a driver without affecting other instances that are servicing separate devices.

DDI/DKI

Device Driver Interface/Device Kernel Interface. These are function call entry points that device drivers should implement in order to fully support DR. DDI/DKI is specified in the “Writing Device Drivers” section of the Driver Developer Site 1.0 AnswerBook (http://docs.sun.com).

DDI_DETACH

A function, used by DR (called from the dr_driver), that provides the ability to detach a particular instance of a driver without affecting other instances that are servicing separate devices.

DDI_RESUME

A function, used by DR (called from the dr_driver), that provides the ability to detach a board that contains the kernel cage (OBP, kernel, and non-pageable memory). The kernel cage can only be relocated after all of the drivers throughout the entire DSD (not just on the board being detached) are quiesced to guarantee the data integrity of the kernel cage relocation. DDI_RESUME resumes the drivers after the quiesce period.

DDI_SUSPEND

A function, used by DR (called from the dr_driver), that provides the ability to detach a board that contains the kernel cage (OBP, kernel, and non-pageable memory). The kernel cage can only be relocated after all of the drivers throughout the entire DSD (not just on the board being detached) are quiesced to guarantee the data integrity of the kernel cage relocation. DDI_SUSPEND suspends the drivers to begin the quiesce period.

decay

The period by which historical usage is discounted.

DEN

The Directory Enabled Networks working group. The goal of this group is to offer a standard information model and directory schemas to tie together users and applications with network elements, protocols, and services through specific relationships. By complying to this information model and the DEN schemas, different network equipment and application vendors should be able to build interoperable network elements around a central directory.

detaching

See [DR Detach]

Diff-Serv

The Differentiated Services working group of the Internet Engineering Task Force (IETF). Diff-Serv addresses network management issues related to end-to-end Quality of Service (QoS) within diverse and complex networks.

DIMM

Dual In-Line memory Module. A memory module with higher capacity and faster performance than SIMMs (Single In-Line Memory Module) It is currently used as the memory source for all Sun Microsystems platforms.

direct control

A means of control that operates on the resource you want to control. For example the Solaris Resource Manager software controls CPU usage per user by implementing a scheduling class that decides who should get what share of the CPU.

DISC

Dynamic internal service classes created by WLM. These classes enable WLM to manage transactions. Each DISC is associated with one or more normal service classes and a given server component. The number of transactions using each route then allow the DISCs to be weighted. Thus, if the external or standard service class goal is not being met, the associated DISCs can be managed (if that is where a bottleneck lies).

distributed queuing system

A batch system product from Florida State University that is available in the public domain. The set of system resources it understands is host (by name), system architecture, operating system type, amount of memory, and CPU usage.

DMTF

Desktop Management Task Force.

DQS

See [distributed queuing system]

DR

See [Dynamic Reconfiguration]

DR Attach

The process of bringing a system board under the control of the Solaris operating environment through use of the DR mechanism.

DR Detach

The process of removing a system board from Solaris operating system control through use of the DR mechanism.

DSD

See [Dynamic System Domains]

DSS

Decision Support System.

DSS/DW

Decision Support System / Data Warehousing.

Dynamic Reconfiguration

(DR) A Sun Microsystems technology supported on the Starfire and other Sun Enterprise servers which allows system boards to be added (attached) or removed (detached) from a single server or domain.

Dynamic System Domains

(DSD) Starfire independent hardware entities formed by the logical association of its system boards. Each domain on the Starfire enjoys complete hardware isolation from other domains, executes its own private version of the Solaris operating system, and is centrally managed by the SSP.

ELIM

See [Extended Load Information Manager]

Enterprise 10000

See [Sun Enterprise 10000]

ERP

Enterprise Resource Planning

error event

A discrete on/off event, as opposed to a continuous variable to be compared against a limit.

exception

A condition that represents a problem in processing a job. LSF can watch for several types of exception conditions during a job's life cycle.

exclusive scheduling

A type of scheduling used by LSF that makes it possible to run exclusive jobs on a host. A job only runs exclusively if it is submitted to an exclusive queue. An exclusive job runs by itself on a host. LSF does not send any other jobs to the host until the exclusive job completes.

Extended Load Information Manager

(ELIM) LSF uses the Load Information Manager (LIM) as its resource monitoring tool. To modify or add load indices, an Extended Load Information Manager can be written.

fairshare

A form of scheduling used by LSF to prevent a single user from using up all the available job slots, thus locking out other users. Fairshare scheduling is an alternative to the default first come, first served scheduling. Fairshare scheduling divides the processing power of the LSF cluster among users and groups to provide fair access to resources for all jobs in a queue. LSF allows fairshare policies to be defined at the queue level so that different queues can have different sharing policies.

FlowAnalyzer (NetFlow)

An application that uses the output from NetFlow FlowCollector. It provides very elaborate processing, graphing, and reporting options that can be used for network analysis, planning, trouble shooting, and more

FlowCollector (NetFlow)

A NetFlow datagram consumer for one or more NetFlow devices. These devices simply point to the host and port number on which the FlowCollector software is running. The FlowCollector aggregates this data, does preprocessing and filtering, and provides several options to save this data to disk (such as flat files). Other applications such as network analyzing, planning, and billing can use these files as input.

Gigaplane-XB

The interconnect on the Starfire that provides main memory access through a point-to-point data router which isolates data traffic between system boards and minimizes any performance degradation when memory interleaving is disabled.

goal

Goal-based policies are prescriptitive rather than reactive. A goal can be translated into a mixture of limits, priorities, and relative importance levels. Goals can include actions to be performed when the goal cannot be met.

Health Monitor

See [SyMON Health Monitor]

heavily damped

A system is heavily damped if you feed back a small proportion of an error over a longer control interval. A heavily damped system tends to be sluggish and unresponsive when a large time constant is used.

hierarchical fairshare

A method of sharing resources, supported by LSF. Hierarchical fairshare enables resources to be allocated to users in a hierarchical manner. Groups of users can collectively be allocated a share, and that share can be further subdivided and given to subgroups, resulting in a share tree.

host based resources

Resources that are not shared among hosts, but are tied to individual hosts. An application must run on that host to access such resources. Examples are CPU, memory, and swap space. Using up these resources on one host does not affect the operation of another host.

Hostview

A GUI program that runs on the SSP machine (which is a component of an Enterprise 10000 system). Hostview enables you to monitor and control the Enterprise 10000. For example, Hostview can display continuous readouts of power and temperature levels at various locations within the Enterprise 10000 server.

HPC

High Performance Computing

HP OpenView

Computer-oriented local and wide area networks are normally managed using SNMP protocols, with Solstice SunNet Manager or HP OpenView products collecting and displaying the data. Both products provide some visibility into what is happening in the computer systems on the network, but they are focused on network topology. Resource management is done on a per-network basis, often by controlling the priority of data flows through intelligent routers and switches.

HTTP

Hypertext Transfer Protocol. HTTP is used by Web servers to host content and respond to HTTP requests from Web browsers.

IBM Workload Manager

A comprehensive tool set for MVS that provides an automated resource management environment, driven by high level business goals, and that, in many cases, is self tuning. Tools are provided to define the business goals or objectives, to control system resources, and to feed metrics concerning these resources back to the resource controller, which attempts to ensure that the goals are met.

IETF

Internet Engineering Task Force.

indirect control

A means of control that works via resources that are dependent upon the resource that is being controlled. For example, to limit the I/O throughput of a process, it is sufficient to be able to measure the I/O throughput and limit the CPU resources for that process.

intercache protocol

(ICP) A protocol used to implement clustered caches.

See also [clustered cache]

interleaving

See [memory interleaving]

intimate shared memory

(ISM) A way of allocating memory so that it can't be paged. The shared memory area is often the largest component of a database's memory requirements, and is the easiest to insulate between database instances. Because intimate shared memory is wired down, the memory allocated to each database instance stays allocated and one instance cannot steal memory from another.

Int-Serv

The Integrated Services working group of the Engineering Task Force (IETF).

IP

Internet Protocol. IP is the foundation of the TCP/IP architecture. It operates on the network layer and supports addressing. IP enables data packets to be routed.

ISM

See [intimate shared memory]

ISP

Internet Service Provider, a company that provides Point-of-Presence access to the Internet.

ISPF

Interactive System Productivity Facility, a generic MVS interface that can be used by the operator/administrator to define, activate, and deactivate policies.

Java Dynamic Management Kit

A JavaBeans based framework for developing and deploying dynamic management based applications. Autonomous agents can be deployed in real-time to perform management tasks for devices on the network.

JTAG

Joint Test Action Group, IEEE Std. 1149.1. JTAG is an alternate communications interface between the SSP machine and the Enterprise 10000 server, and is used when the standard network connection between the SSP and the Enterprise 10000 is unavailable.

Jiro

A technology being developed at Sun to address modern storage issues. Storage is now open for access in a heterogeneous multivendor environment, where multiple server and storage vendors can all be connected over the Storage Area Network (SAN). This is an emerging technology, and tools to manage a SAN are still being developed. Jiro is based on a distributed pure Java framework that can run on servers from any vendor, interface to other storage management software, and manage any kind of attached storage.

Java Virtual Machine

The machine image, implemented in software, upon which Java code runs.

JVM

See [Java Virtual Machine]

kernel cage

A special data structure (normally contained within a single system board) that controls the dynamic growth of all non-relocatable memory, including the OpenBoot PROM (OBP) and kernel memory. When Dynamic Reconfiguration (DR) is used to detach a system board containing the kernel cage, it is necessary to quiesce the operating system to ensure that no I/O or kernel activity occurs while the kernel cage is being relocated.

kernel memory

Memory that is used to run the operating system.

kernel module

A Solaris Bandwidth Manager module that contains the classifier and the scheduler.

LAN

See [local area network]

lightly damped

If you feed back a large proportion of an error with a short control interval, the system is said to be lightly damped. A lightly damped system is very responsive to sudden changes but will probably oscillate back and forth.

LIM

See [Load Information Manager]

limit

A simple rule with a single input measurement. It is also common to have several thresholds with a warning level action and a critical problem level action for the same measure.

lnode

Limit node, a node in a special resource tree used by Solaris Resource Manager (SRM). SRM is built around lnodes which are a fundamental addition to the Solaris kernel. lnodes correspond to UNIX UIDs, and may represent individual users, groups of users, applications, and special requirements. The lnodes are indexed by UID and are used to record resource allocations policies and accrued resource usage data by processes at the user, group of users, and application level.

Load Information Manager

(LIM) The resource monitoring tool used by LSF. The Load Information Manager process running on each execution host is responsible for collecting load information. The load indices that are collected include: host status, length of run queue, CPU utilization, paging activity, available swap space, available memory, and I/O activity.

Load Share Facility

(LSF) A software facility that provides the capability of executing batch and interactive jobs on a pool of networked computers. The Sun Microsystems High Performance Computing (HPC) package includes the Load Share Facility as a vehicle for launching parallel applications on an HPC cluster. In addition to starting batch jobs, the Load Share Facility also provides load balancing.

local area network

A set of computer systems in relatively close proximity that can communicate by means of networking hardware and software.

LPAR

Logical Partitions, an IBM S/390™ logical entity which runs its own operating system instance and it's allocated resources and managed by PR/SM.

LSF

See [Load Share Facility]

LWP

Lightweight Process

Management Information Base

A database that contains network management variables and can be accessed via SNMP.

master host

The node where the LSF batch queues reside. When the LSF software initializes, one of the nodes in the cluster is elected to be the master host. This election is based on the order of nodes listed in a configuration file. If the first node listed in the configuration file is inoperative, the next node is chosen, and so forth.

maximum bandwidth

The amount of spare bandwidth allocated to a class by the Solaris Bandwidth Manager. The maximum bandwidth is dependent upon the percentage of bandwidth the class can borrow.

MDF

Multiple Domain Facility, an Amdahl Corporation™ technology which provides logical partitioning for its mainframes. By integrating special hardware for each logical partition or domain, Amdahl processor complexes could run multiple operating systems at close to native performance.

memory interleaving

A method of using computer memory that helps increase memory subsystem performance by reducing the probability of hot spots or contention in a few memory banks. This is accomplished by spreading access to multiple memory banks.

Message Passing Interface

(MPI) An industry standard interface used to parallelize applications.

MIB

See [Management Information Base]

microstate accounting

A method of accounting for resource usage where a high-resolution timestamp is taken on every state change, every system call, every page fault, and every scheduler change. Microstate accounting provides much greater accuracy than sampled measurements.

MPI

See [Message Passing Interface]

MTS

See [Multi-threaded mode]

Multi-threaded mode

A database topology where a single process serves many users.

negative feedback

A method of applying feedback to a system where you take the error difference between what you wanted and what you got, and apply the inverse of the error to the system to reduce the error in the future.

NetFlow

A product from Cisco that is supported by Solaris Bandwidth Manager. NetFlow allows for detailed network measurements that can be sent to other software packages which can process and analyze the data.

Network File System

An application that utilizes TCP/IP to provide distributed file services.

network queuing system

(NQS) A public domain software product that has been enhanced by many hardware vendors. Sterling Software offers a distributed version of NQS called NQS/Exec which is geared toward a supercomputer environment. Limited load balancing is provided as there is no concept of demand queues, since it uses traditional push queues instead. There is also no control over interactive batch jobs.

NFS

See [Network File System]

NQS

See [network queuing system]

NVRAM

Non-Volatile Random Access Memory

OBP

OpenBoot PROM

ODS

Informix OnLine Dynamic Server

OLTP

Online Transaction Processing

operational policy

A policy that is implemented manually as part of operations management. For example, an availability policy can include a goal for uptime and an automatic way to measure and report the uptime over a period. There is no direct control in the system that affects uptime. It is handled by operations staff.

Oracle8i

An Oracle facility that ensures system resources are applied to the most important tasks of the enterprise at the levels required to meet enterprise goals.

PC NetLink

A product from Sun Microsystems that is based on the AT&T Advanced Server for UNIX. PC NetLink adds functionality that was not previously available on Solaris servers with products such Samba and SunLink™ PC™ (a.k.a. Syntax TotalNET Advanced Server). PC NetLink adds file and print services, and enables Solaris servers to act as Microsoft® Windows NT™ Primary Domain Controllers (PDC) or Backup Domain Controllers (PDC). For enterprises with mixed NT and Solaris servers and desktops, PC NetLink 1.0 offers many new options for utilizing hardware resources and minimizing system administration overhead.

Platform Computing Load Share Facility

See [Load Share Facility]

PDP

See [policy decision point]

PEP

See [policy enforcement point]

performance index

The ratio of work completed versus the amount of work which should have been completed in order to meet the goal.

PIN

See [policy ignorant node]

policy agent

A component of the Solaris Bandwidth Manager that implements the configuration and handles communication with the kernel module.

policy control

The application of rules to determine whether or not access to a particular resource should be granted.

policy decision point

In policy administration, the point where policy decisions are made.

policy element

A subdivision of policy objects. A policy element contains single units of information necessary for the evaluation of policy rules. Examples of policy elements include the identity of the requesting user or application, user or application credentials, and so forth. The policy elements themselves are expected to be independent of which Quality of Service signaling protocol is used.

policy enforcement point

In policy administration, the point where policy decisions are enforced.

policy ignorant node

A network element that does not explicitly support policy control using the mechanisms defined in the applicable standard policy.

policy object

An object that contains policy-related information, such as policy elements, and is carried in a request or response related to resource allocation decisions.

policy protocol

A protocol for communication between the policy decision point and policy enforcement point. The policy protocol can be any combination of COPS, SNMP, and Telnet/CLI.

POST

Power-ON self tests, a suite of hardware diagnostic tests which ensure full functionality of a system board.

preemptive scheduling

A method of scheduling where a high priority job can bump a lower priority job that is currently running. LSF provides several resource controls to prioritize the order in which batch jobs are run. Batch jobs can be scheduled to run on a first come first served basis, fair sharing between all batch jobs, and preemptive scheduling.

priority

A relative importance level that can be given to the work done by a system as part of a policy that prioritizes some activities over others.

priority decay

See [process priority decay]

priority paging

A method of implementing a memory policy with different importance factors for different memory types. Application memory is allocated at a higher priority than file system memory, which prevents the file system from stealing memory from other applications. Priority paging is implemented in the Solaris 7 operating environment.

process measurements

Measurements that show the activity of each user and each application.

process memory

Memory allocated to processes and applications.

Process Monitor

An optional module within Sun Enterprise SyMON that can be used to view all the processes on a system. The Process Monitor can also be configured to pattern match and accumulate all the processes that make up a workload.

process priority decay

A process decay method used by SRM, where each processes priority is decayed according to a fixed decay factor at regular intervals (each second).

processor reservation

A method that allows job slots to be reserved for a parallel job until enough are available to start the job. When a job slot is reserved for a job, it is unavailable to other jobs. Processor reservation helps to ensure that large parallel jobs are able to run without under utilizing resources.

processor set

The set of processors available to a system.

provider DSD

Dynamic Reconfiguration (DR) on the Starfire allows the logical detachment of a system board from a provider DSD (the DSD from which resources are borrowed) and the logical attachment of the same system board to a receptor DSD (the DSD where loaned resources are applied).

provider domain

When relocating resources between DSDs, a “provider domain” is the domain where a system board gets logically detached from to then have it attached to a “receptor domain”.

proxy cache

A method of caching Web pages. A proxy caching Web server sits between a large number of users and the Internet, funneling all activity through the cache. Proxy caches are used in corporate intranets and at ISPs. When all the users are active at once, regardless of where they are connecting to, the proxy cache server will get very busy

PR/SM

Processor Resource/Systems Manager), an IBM S/390 hardware feature which allows customers to statically allocate processor and I/O resources to LPARs to concurrently run multiple operating system instances on the same machine.

QoS

See [Quality of Service]

Quality of Service

A measure of the speed and reliability of a service. Solaris Bandwidth Manager provides the means to manage your network resources to provide Quality of Service to network users. QoS is a network-wide issue; if congestion takes place anywhere on the network, it affects the overall quality of service.

RAS

Reliability, Accessibility and Serviceability

receptor DSD

Dynamic Reconfiguration (DR) on the Starfire allows the logical detachment of a system board from a provider DSD (the DSD from which resources are borrowed) and the logical attachment of the same system board to a receptor DSD (the DSD where loaned resources are applied).

receptor domain

When relocating resources between DSDs, a “receptor domain” is the domain which receives a system board after having it logically detached from a “provider domain.”

repository access protocol

The protocol used to communicate between a policy repository and the repository client. LDAP is one example of a repository access protocol.

Resource Management Facility

A component of WLM that tracks metrics including progress against goals.

RMF

See [Resource Management Facility]

RSVP

A protocol (part of the Int-Serv framework), that provides applications the ability to have multiple levels of Quality of Service (QoS) when delivering data across the network. RSVP provides a way for an application to communicate its desired level of service to the network components. It requires each hop from end-to-end be RSVP-enabled, including the application itself (through an API). Bandwidth is reserved at each hop along the way before transmitting begins, guaranteeing that enough resources will be available for the duration of the connection.

SAN

See [Storage Area Network]

scheduler

A component of the Solaris Resource Manager (SRM) that schedules users and applications.

scheduler term

The period of time during which the Solaris Resource Manager (SRM) ensures that a particular user or application receives its fair share of resources.

security policy

A type of policy that aims at preventing access to certain resources or allowing designated users to manage subsystems. For example, Sun Enterprise SyMON 2.0 software includes access control lists for operations that change the state of a system, and multiple network domain views to give different administrative roles their own view of the resources being managed.

SE Toolkit

A toolkit that can be used to develop customized process monitors. The Solaris software can provide a great deal of per-process information that is not collected and displayed by the ps command or Sun Enterprise SyMON 2.0 software. The data can be viewed and processed by a custom written process monitor. You could write one from scratch or use the experimental scripts provided as part of the SE Toolkit. The SE Toolkit is a freely available but unsupported product for Solaris systems. It can be downloaded from the http://www.sun.com/sun-on-net/performance/se3.

server consolidation

A current trend by data centers to reduce cost of server ownership by reducing physical footprint and reducing number and management cost of multivendor platforms. The basis of server consolidation is to combine applications and data contained in several smaller servers into a single larger server.

service class

A class that defines a set of goals, together with periods, duration, and importance. A number of individual processes and CICS/IMS transactions can be assigned membership of a service class. They will then become subject to the specified goals and constraints, including those imposed by any resource group subscribed to by the class. In essence, this is analogous to the SRM lnode, which effectively defines a resource management policy that can be subscribed to.

Service Level Agreement

A written agreement between system managers and end users that captures the expectations and interactions between end users, system managers, vendors, and computer systems. Often, many additional interactions and assumptions are not captured formally.

Service Level Management

The process by which information technology (IT) infrastructure is planned, designed, and implemented to provide the levels of functionality, performance, and availability required to meet business or organizational demands.

service provider

In a network policy, the service provider controls the network infrastructure and may be responsible for the charging and accounting of services.

service time

The time it takes for an I/O device to service a request. This can be complex to measure. For example, with today's disk storage systems, the device driver issues a request, that request is queued internally by the RAID controller and the disk drive, and several more requests can be sent before the first one comes back. The service time, as measured by the device driver, varies according to the load level and queue length and is not directly comparable to the old-style service time of a simple disk drive.

SEVM

Sun Enterprise Volume Manager, technically equivalent to Veritas Volume Manager.

ShareII

A resource management product from product from Softway. The Solaris Resource Manager (SRM) is based on ShareII.

shared resources

A resource that is not tied to a specific host, but is associated with the entire cluster, or a specific subset of hosts within the cluster. Examples of shared resources include: floating licenses for software packages, disk space on a file server which is mounted by several machines, and the physical network connecting the hosts.

SHR Scheduler

A component of the Solaris Resource Manager (SRM) that controls the CPU resources. Users are dynamically allocated CPU time in proportion to the number of shares they possess (analogous to shares in a company), and in inverse proportion to their recent usage. The important feature of the SHR scheduler is that while it manages the scheduling of individual threads, it also portions CPU resources between users.

Simple Network Management Protocol

(SNMP) An open network protocol used by network management systems that are based on TCP/IP.

SLA

See [Service Level Agreement]

SNIA

Storage Network Industry Association.

SNMP

See [Simple Network Management Protocol]

Solaris Bandwidth Manager

A product from Sun that provides the means to manage your network resources to provide Quality of Service (QoS) to network users. It allows network traffic to be allocated to separate Classes of Service (CoS), so that urgent traffic gets higher priority than less important traffic. Different classes of service can be guaranteed a portion of the network bandwidth, leading to more predictable network loads and overall system behavior. Service Level Agreements can be defined and translated into Solaris Bandwidth Manager controls and policies. Tools and APIs provide an interface for monitoring, billing, and accounting options.

Solaris Management Console

An application that provides a generic framework for gathering together operating system administration tools and interfacing to industry standard initiatives such as the Web-based management initiative (WebM) and the Common Information Model (CIM).

Solaris Resource Manager

(SRM) A software tool for enabling resource availability for users, groups, and applications. The Solaris Resource Manager provides the ability to allocate and control major system resources such as CPU, virtual memory, and number of processes. The Solaris Resource Manager software is the key enabler for server consolidation and increased system resource utilization.

Solstice SunNet Manager

Computer-oriented local and wide area networks are normally managed using SNMP protocols, with Solstice SunNet Manager or HP OpenView products collecting and displaying the data. Both products provide some visibility into what is happening in the computer systems on the network, but they are focused on network topology. Resource management is done on a per-network basis, often by controlling the priority of data flows through intelligent routers and switches.

SPARCcluster

A highly integrated product line that is focused on improved availability in commercial environments. Its management tools will eventually become an integrated extension to the Sun Enterprise SyMON2.0 software. For High Performance Computing, Sun HPC Servers use the Platform Computing Load Share Facility (LSF) to perform load balancing on much larger and more loosely coupled clusters.

SRM

See [Solaris Resource Manager]

SRM(IBM)

The System Resource Manager of WLM. The term SRM(IBM) is used in this book to differentiate it from Solaris Resource Manager. SRM(IBM) provides the algorithms for managing resources and caters for dynamic switching between compatibility and goal modes.

SSP

System Service Processor. Starfire's system administrator & system monitoring interface. The SSP configures the Starfire hardware, through a private ethernet link, to create domains. The SSP collects hardware logs, provides boot functions, and produces consoles for each domain.

Starfire

See [Sun Enterprise 10000]

static resources

Host information that does not change over time, such as the maximum RAM available to processes running on the host.

Storage Area Network

(SAN) A complex managed storage system, where networked storage using fiber channel makes up an interconnection layer between multiple servers or clusters and multiple storage subsystems. A storage area network can contain switches and routers just like local or wide area networks, but the protocol in common use is SCSI over fiber channel rather than IP over ethernet. A storage area network may also span multiple sites, for example where remote mirroring is being used for disaster recovery.

submission host

In a typical LSF workload configuration, the submission host is the node where the user or operator submits the task to be performed.

Sun Enterprise 10000

A highly scalable 64-processor (UltraSparc II) SMP server with up to 64 Gbytes of memory and over 20 Tbytes of disk space.

Sun Enterprise SyMON 2.0

A product developed by Sun to act as a user interface to hardware features. It is a powerful and extensible system and network monitoring platform that is used to manage other products. Sun Enterprise SyMON 2.0 is a Java-based monitor with multiple user consoles that can monitor multiple systems using the secure extensions to SNMPv2 to communicate over the network.

SunNet Manager

See [Solstice SunNet Manager]

SyMON

See [Sun Enterprise SyMON 2.0]

SyMON Health Monitor

A SyMON module that can be used in a resource management scenario to determine if a system has enough resources to run comfortably. For example, if the CPU state is reported as “red”, then either less work or more CPU power may be needed on that system. Similarly, if the memory rule reports “red” then the system may need more memory

system level measurements

A type of measurement. System level measurements show the basic activity and utilization of the memory system and CPUs. Some network measurements such as TCP/IP throughput are also available on a per system basis. Per process activity can be aggregated at a per system level then combined with network measurements to measure distributed applications.

Teamquest

A workload analysis product. See www.teamquest.com

time constant

In control theory, the rate at which a system responds to changes.

TNF

See [trace normal form]

ToS

See [Type of Service]

trace normal form

(TNF) A format used to implement tracing (which makes it possible to trace the execution steps of user and kernel processes). Trace normal form, which is supported by the Solaris operating environment, provides a self-describing trace output format. Trace normal form allows data structures to be embedded in the trace file without the need for an external definition of their types and contents.

Type of Service

(ToS) A header field contained in IP packets. Its purpose is to convey information about how the packet should be routed. The Solaris Bandwidth Manager can use this information when classifying a packet. It can also change the information, to influence how the packet is routed.

UDB

DB2 Universal Database.

usage decay

A form of decay used by SRM. The user scheduler is the most important and visible portion of SRM and it implements usage decays which control long term CPU allocation responsiveness.

virtual memory

A type of memory that is allocated from a central resource pool and is consumed by an application when it requests memory from the operating system. Virtual memory is not directly related to physical memory usage because virtual memory is not always associated with physical memory. For example, if an application requests 16 Mbytes from the operating system, the operating system will create 16 Mbytes of memory within that application's address space but will not allocate physical memory to it until that memory is read from or written to.

virtual Web hosting

A web server configuration where a single server is configured to respond to hundreds or thousands of Internet addresses. Virtual web hosting is often used in situations where web sites receive little or no activity most of the time. In these situations, it is usually too expensive to dedicate a single computer system to each web site.

WAN

See [wide area network]

WebM

Web-based management initiative

wide area network

A network that provides connectivity across a large geographical area.

WLM

See [IBM Workload Manager]

Workload Manager

See [IBM Workload Manager]

XCF

See [Cross-System Coupling Facility]