Acronyms and Glossary
- 1 Acronyms
- 2 Glossary
- 2.1 adaptive
- 2.2 application server
- 2.3 Arc13 or Arc14
- 2.4 bandwidth
- 2.5 batch process
- 2.6 batch productivity
- 2.7 benchmark
- 2.8 casual user
- 2.9 capacity
- 2.10 capacity planning
- 2.11 Capacity Planning Tool (CPT)
- 2.12 client
- 2.13 complexity
- 2.14 component service time
- 2.15 composite workflow
- 2.16 compute-intensive
- 2.17 compute server
- 2.18 computer processor chip
- 2.19 concurrent user
- 2.20 ConfigStore
- 2.21 container machine
- 2.22 core
- 2.23 data center
- 2.24 data server
- 2.25 data source
- 2.26 • SDE DC/DBMS.
- 2.27 • SDE DBMS.
- 2.28 • Small shape file.
- 2.29 • Large shapefile.
- 2.30 • File geodatabase.
- 2.31 display or workflow display transaction
- 2.32 display complexity
- 2.33 display cycle time
- 2.34 display queue time
- 2.35 display response time
- 2.36 display service time
- 2.37 distributed
- 2.38 federated ArcGIS for Server site
- 2.39 file server
- 2.40 floating-point
- 2.41 geodatabase
- 2.42 geodatabase server
- 2.43 geodatabase transition
- 2.44 GIS Server Site
- 2.45 GIS Server Cluster
- 2.46 high-available configuration
- 2.47 hosted ArcGIS for Server site
- 2.48 I/O intensive
- 2.49 image data
- 2.50 information product
- 2.51 integer
- 2.52 Internet
- 2.53 local area network (LAN)
- 2.54 local site
- 2.55 map server
- 2.56 multicore chips
- 2.57 multithread process
- 2.58 network
- 2.59 network bandwidth
- 2.60 network capacity
- 2.61 network interface card (NIC)
- 2.62 network queue time.
- 2.63 network response time
- 2.64 display response time
- 2.65 network service time
- 2.66 network throughput
- 2.67 network traffic
- 2.68 network workflow traffic
- 2.69 operations
- 2.70 parallel processing
- 2.71 peak workflow throughput
- 2.72 performance
- 2.73 performance baseline
- 2.74 performance benchmark
- 2.75 performance budgets
- 2.76 performance model
- 2.77 performance targets
- 2.78 platform architecture
- 2.79 platform capacity
- 2.80 platform configuration
- 2.81 platform load
- 2.82 platform node
- 2.83 platform operating system
- 2.84 platform service times
- 2.85 platform sizing
- 2.86 platform sizing chart
- 2.87 platform tier
- 2.88 platform vendor
- 2.89 process queue time
- 2.90 processing
- 2.91 processor chip
- 2.92 processor core
- 2.93 processor MHz
- 2.94 processor socket
- 2.95 productivity
- 2.96 queuing theory
- 2.97 queue time
- 2.98 random arrival times
- 2.99 registered service
- 2.100 relative performance
- 2.101 remote desktop services
- 2.102 remote site
- 2.103 requirements analysis
- 2.104 response time
- 2.105 sequential processing
- 2.106 server
- 2.107 service provider
- 2.108 service time
- 2.109 single-thread processing
- 2.110 site bandwidth
- 2.111 software baseline
- 2.112 software service times
- 2.113 SPEC baseline performance value
- 2.114 SRint2006
- 2.115 standard ESRI workflows
- 2.116 storage
- 2.117 stovepipe
- 2.118 terminal server
- 2.119 throughput
- 2.120 transaction
- 2.121 user productivity
- 2.122 user think time
- 2.123 user workflow
- 2.124 utility services
- 2.125 utilization
- 2.126 Web application server (WA)
- 2.127 Web browser
- 2.128 Web server
- 2.129 wide area network (WAN)
- 2.130 Windows Terminal Server (WTS)
- 2.131 work transaction
- 2.132 workflow capacity
- 2.133 workflow performance
- 2.134 workflow performance summary
- 2.135 workflow requirements
- 2.136 workflow service times
- 2.137 workstation
ADF Application Development Framework
AML ARC Macro Language
API application program interface
ASC application server connect
ASP application service provider
CIFS Common Internet File Services
COTS commercial off-the-shelf
CPT capacity planning tool
DC direct connect
DCOM Distributed Component Object Model
DMZ demilitarized zone
DPM displays per minute
DS data source
EDN ESRI Developer Network
Gbps gigabits per second
g.net (regional) geography network
GIS geographic information system
HTTP Hypertext Transfer Protocol
ICA independent computing architecture
IP Internet Protocol
IPD Information Product
ISP Internet service provider
JBOD just a bunch of disks
Kbps kilobits per second
LAN local area network
MAC Media Access Control
Mbps megabits per second
NFS network file server
NIC Network Interface Card
ODBC Open Database Connectivity
RAID redundant array of inexpensive disks
RDP remote desktop protocol
RDS remote desktop services
RDSH Remote Desktop Services Host
SAN storage area network
SDE Spatial Database Engine
SM service manager
SMP Symmetrical Multiple Processor
SOA service-oriented architecture
SOAP Simple Object Access Protocol
SOC Server Object Containers
SOM Service Object Manager
SPEC Standard Performance Evaluation Corporation
SS spatial services
SSL Secure Socket Layer
SSO single sign on
TCP Transmission Control Protocol
TPH Transactions per hour
WA Web applications
WAN wide area network
WSE Web services extensions
WTS Windows Terminal Server
XML Extensible Markup Language
In the context of system design, a term that describes something that helps you or your organization adapt to technology change; for example, an architecture that can increase capacity by adding more servers is adaptive to increasing system capacity needs.
A computer platform used to host user application software executables that are managed and deployed from a central computer facility. These servers can include desktop user application sessions (Windows Terminal Servers) or Web user transactions (Web servers, map servers, server object container machines).
Arc13 or Arc14
See performance baseline.
A data rate measured in bits per sec.; also referred to as peak network throughput. The digital data rate limit (or network capacity) of a physical communication link is related to its bandwidth in hertz.
A computer program (procedure) or series of procedures executed without user interaction. A batch process would normally execute for a period of time significantly beyond that of an average user display transaction.
A measure of the workflow processing throughput of a batch process expressed in terms of peak displays per minute (DPM).
See performance benchmark.
Type of user workflow with productivity less than a power user. Specific user productivity for each workflow can be adjusted on the CPT workflow tab, or modified on the CPT Design or Calculator tab to represent different user productivity rates.
The peak computer platform or network processing throughput. Capacity can refer to a single component or an integrated system of platform and network components. The peak system throughput is limited by the capacity of the weakest system component.
A process of defining individual component performance and capacity specifications based on estimated peak user workflow requirements. Capacity planning is about building a distributed hardware and network infrastructure to support enterprise GIS operations.
Capacity Planning Tool (CPT)
The Microsoft Office Excel workbook application provided with this book. Readers of this book can use the CPT to collect peak user workflow requirements and establish their own specific target performance and capacity specifications to ensure successful GIS deployment.
The term used to describe the location of the user computer display.
see display complexity.
component service time
The total program execution or network transfer time required for each system component to complete processing of an average work transaction. Component service times are normally discussed in reference to an integrated system of components, where the sum of the component service times plus any additional processing delay or wait times are combined to represent the work transaction response time.
Many Web applications today are developed from a group of published Web service components. A custom user workflow created by combining two or more single user workflow software service times is referred to as a composite workflow.
Adjective describing a computer software process that consumes a significant amount of server processing resources; a process whose peak throughput is determined by the processing capacity of the platform CPU.
An IT term used to describe a server located in the data center that is used to host primary application executables; used to distinguish compute servers from data servers in the computer room. Compute is an adjective describing the type of computer or server platform or environment. (In early GIS history, we used “compute server” when referring to UNIX servers that hosted ARC/INFO applications serving terminal clients. We could not use “terminal server” in those days, since that term was used to identify wall network connections for mainframe terminals. Terms have evolved over the years. The mainframe wall network connections have gone away, so this type of server can now be called a “terminal server.”)
computer processor chip
The small integrated circuit board that supports the computer core processors. Commodity Intel computers normally contain two computer processor chips, with each chip supporting one, two, or four processor cores. computer network A group of interconnected computers. Computer network interface cards (NIC) connect computers to network switches that are used as a connection point for groups of computers and supporting network devices (printers, plotters, etc.). Network capacity is established by the capacity of the switch connections (ports).
A term used in capacity planning to identify the number of concurrent workflow users that can be supported on a baseline platform performance. Concurrent users can be translated to transactions per minute by multiplying by user productivity. Peak transaction rates are used to represent system loads (i.e. 100 concurrent users x 6 displays per minute/user = 600 displays per minute). Peak transaction rates are normally measured during the peak hour of the day and then divided by 60 to identify an average transactions per minute. The workflow service times and traffic per display provide the estimated system loads for each single transaction. It is assumed that these transaction loads are distributed randomly for modeling purposes.
When ESRI first delved into capacity planning, a simple capacity relationship was used along with published vendor SPEC benchmarks to translate peak concurrent user loads to other platform environments. Performance baselines and models were updated on an annual basis to address changes in technology. As the models matured, concurrent users were were translated to peak transaction rates for modeling purposes.
ArcGIS for Server ConfigStore was introduced with the ArcGIS 10.1 release. ConfigStore maintains service configurations and cluster assignments for machines deployed within a single GIS Server Site.
The server object container (SOC). A term used with ArcGIS Server describing the machine hosting the software (ArcObjects) and deployed SOC instances.
See processor core.
A central computer facility used to house centrally managed servers and computer data resources. Data centers employ enhanced security and system management software to administer enterprise applications and database server resources, thus providing services to client computers located across the local area network (LAN). The data center is also a central network access point for wide area network (WAN) and Internet user locations.
Server platforms that host database software or provide network access to a file system repository.
The term used with the Capacity Planning Tool to identify the source of distributed data resources. The following shorthand notation is used to represent standard GIS workflow data sources:
• SDE DC/DBMS.
ArcGIS applications connecting to a geodatabase using an ArcGIS direct-connect communication architecture.
====• SDE server/DBMS====. ArcGIS applications connecting to a middle-tier ArcSDE remote server, which then connects to a separate geodatabase server.
• SDE DBMS.
ArcGIS applications connecting to a geodatabase using an ArcGIS SDE communication architecture, with ArcSDE installed on the DBMS server.
• Small shape file.
Term used for a shape file data source with the extent roughly the size of a city neighborhood (9 vector layers, 145,000 features, 140 MB with additional 241 MB raster layer).
• Large shapefile.
Term used for a shapefile data source with the extent roughly the size of a large city (9 vector layers, 10,000,000 features, 1.2 GB with additional 241 MB raster layer).
• File geodatabase.
ArcGIS file-based XML data source supporting geodatabase schema and up to a terabyte of geospatial data.
display or workflow display transaction
Refers to a unit of work for the purpose of measuring a user workflow throughput (work transaction). A display can refer to an average map display transaction (i.e., Web map transaction, or Web service that returns a map) or a unit of processing time (i.e., average processing loads generated by a power user workflow during a 6-second display cycle time). One way to define an average workflow display would be to collect the CPU processing and network traffic throughput generated by a single power user performing a workflow over a 10-minute period and divide the results by 100.
Term used to identify the relative service time required to process an average workflow display or transaction. The CPT identifies a medium complexity baseline processing load for the most popular GIS workflows. Light complexity workflows generate half the baseline processing load. Heavy complexity workflows generate fifty percent more that the baseline processing loads.
display cycle time
The total time between client workflow displays; it is equal to the display response time plus the user think time.
display queue time
Wait time. Any process delay times due to use of shared system resources. Wait times result when component capacity exceeds 50 percent utilization due to random transaction arrival times plus any cumulative travel times identified as system latency.
display response time
See also network display response time. The total amount of time from user display request until the complete display is returned to the user interface. Display response time includes all system component service times plus all queue times.
display service time
The total program execution or network transfer time required for a system component (platform or network) to complete processing or transfer of an average work unit.
Used here to describe operations or system architectures that include shared system components such as networks, application servers (Windows Terminal Server or Web server farms), and a common data source, or geodatabase in GIS parlance. A distributed database allows two or more people to be working on the same data in separate locations. (Capacity planning is about building a distributed hardware and network infrastructure to support enterprise GIS operations.)
federated ArcGIS for Server site
The term used with Portal for ArcGIS for identifying ArcGIS for Server sites that share a common portal user store. The Portal for ArcGIS provides a single sign-on experience for all federated ArcGIS for Server sites. Any items published to a federated server will automatically be shared on the portal. The Portal becomes the token generator for user service access to a federated ArcGIS for Server site. Several ArcGIS for Server sites can be registered with a single Portal for ArcGIS.
Server platforms that provide network access to a file system repository.
Floating-point numbers are the set of real numbers that include all negative and positive values, wherein fractional values are represented as decimal numbers. Real numbers are represented by the computer with a fixed field length. The decimal point within the numeric representation is allowed to float to take advantage of the full field length. The accuracy of the real number will depend on the number of digits in the field length and the size of the number (larger numbers will have reduced accuracy due to the fixed number of digits in the numerical representation). The most common floating-point representation in computers is that defined by the IEEE 754 standard.
A database designed to store, query, and manipulate geographic information and spatial data. It is also known as a spatial database. Within a spatial database, spatial data is treated as any other data type. Vector data can be stored as point, line, or polygon data types, and may have an associated spatial reference system. A geodatabase record can use a geometry data type to represent the location of an object in the physical world and other standard database data types to store the object’s associated attributes. Some geodatabase formats, such as those used by ESRI in its ArcGIS software, also include support for storing raster data.
The server platform hosting the geodatabase.
A term used to distinguish between database replication and geodatabase replication.
GIS Server Site
ArcGIS for Server install sharing a common ConfigStore. ConfigStore was introduced with the ArcGIS 10.1 release. GIS Server Site can include multiple clusters. Each cluster can include multiple machines, with each machine within a cluster deploying the same service configurations.
GIS Server Cluster
ArcGIS for Server Site can include multiple Clusters, with machines deployed in the site assigned to a specific cluster. GIS Server clusters were introduced with the ArcGIS 10.1 release. All machines within a site cluster will deploy the same service configuration.
One that continues to perform as a system despite the loss of any individual server. See chapter 7 for high-available Web configurations.
hosted ArcGIS for Server site
The term used with Portal for ArcGIS for identifying the ArcGIS for Server site that hosts Portal services. The host server must first be a federated server with the Portal for ArcGIS site. The designated ArcGIS for Server site will host the tiled map services and feature services published on the Portal. Portal for ArcGIS can have only one hosted ArcGIS for Server site. The hosted services (feature or tile) are created under a new Hosted folder in the host ArcGIS for Server configuration store.
Refers to a workflow process where peak throughput is limited by the system traffic throughput capacity (input/output) of the system environment. System traffic throughput capacity is a combination of the throughput capacities of both platform traffic and the network. The platform traffic throughput capacity is normally limited by the installed network interface card (NIC). The network throughput capacity is determined by the most limited shared bandwidth segment (network router or switch interface).
Raster or picture data types, where a single or group of picture files are assembled to represent a data layer.
Roger Tomlinson’s term for data transformed by GIS into information particularly useful to you or your organization, and delivered via computer, often but not always in the visual form of a map. An example is economic data analyzed in relation to a specific location. See Tomlinson’s Thinking About GIS.
Integer numbers are the set of values (0, 1, 2, 3, etc.) and their negative values (0, -1, -2, -3, etc.) that are represented without a decimal point. Integer numbers represented in a computer with a fixed field length, where the number of significant digits to the right of the decimal point are a fixed value. The accuracy of an integer number will be constant based on the number of significant decimal values. The maximum size of the number will be limited by the fixed field length.
The world wide public communication network. User access to the public Internet infrastructure is provided by private Internet service providers (ISPs) that charge for the service connection.
local area network (LAN)
Private network infrastructures connecting computers within a relatively small area, such as a building or a campus. Local networks are supported by switched Ethernet protocols that enable high bandwidth communications between computers and shared network devices (printers, plotters, etc.) within the building or group of buildings.
A site that supports computers connected by a LAN. Connections can include campus networks that have several LAN environments connected together over shared switched network connections or network routers, using common LAN protocols.
A computer platform used to host Web mapping server software. This was a common term used in reference to hosting early Web mapping software. See container machine.
A computer chip with more than one processor core manufactured on a single chip die. The multicore chips share a level 2 (L2) cache that improves processing performance by reducing time to access program instructions. Multicore chip designs were introduced by computer hardware vendors in the 2004–2006 timeframe, in an effort to continue improving computer performance and capacity while reducing hardware costs. Current AMD and Intel processors provide two- and four-core chip designs, with higher capacity designs (8, 16, etc.) promised in future technology releases. Sun T-series processors already contain eight core processors per chip, and IBM Power PC and HP Itanium processors provide dual-core chips.
A popular software execution model that allows multiple threads (independent services) to exist within the context of a single process, sharing the software code resources while providing independent execution. Multithreaded software reduces memory consumption by running the same number of service threads within a reduced number of processes.
See computer network.
Network capacity is defined by the capacity (bandwidth) of the router or switch connecting each network component to the client-to-server communication path.
network interface card (NIC)
The network interface card is a network component device used to connect computers to the physical local area network (LAN).
network queue time.
Network wait time or display queue time, all three variations are used to identify any transport delay times due to use of shared system resources. Wait times result when network traffic throughput exceeds 50 percent of the available network bandwidth capacity due to random transaction arrival times and any cumulative travel times identified as system latency.
network response time
Term used to identify the total amount of network transfer time from initial data request until the data transfer reaches the user interface.
display response time
Term used to identify the total amount of time from initial data request until the final display is rendered on the client computer. Display response time includes display processing time, network transport time, and any system delays due to service contention.
network service time
The term network service time is use to identify the data transfer time of an average work unit (workflow display) over the limiting network connection bandwidth. Estimated data transport time can be calculated by dividing the total traffic per display by the limiting network bandwidth connection.
The amount of digital data per time unit that is delivered over a physical or logical link, or that is passing through a certain network node. For example, it may be the amount of data that is delivered to a certain network terminal or host computer, or between two specific computers. The throughput is usually measured in bits per second (bits/s or bps).
A common term for network throughput. Traffic throughput is measured in thousands (kilo), millions (mega), or billions (giga) bits per second.
network workflow traffic
Includes client traffic or database traffic, meaning the data transfer volume of an average work unit (traffic per display). Client traffic per display, measured in Mbpd, is a measure of communication between the client and central data center, and is used to compute client network transport service times. The database (DB) traffic per display, measured in Mbpd, is a measurement of traffic between the servers in the data center, and is used in calculating how much server NIC traffic to anticipate (all server traffic flows through the server NIC interface).
A general term synonymous with “systems”; in the context of system design in particular, it refers to the well-defined user workflows (processes and procedures) in place, underpinning the daily work efforts of an organization. Oil and gas companies have used the term with GIS since the early 1990s; within local governments, the term is used for enterprise applications (permit systems, customer service systems, accounting systems).
The simultaneous use of more than one processor core to execute a single program. Most computer programs are executed sequentially, one instruction after the next, just as most human procedures are completed as a sequential series of procedural steps. Parallel processing requires that several instructions of a single procedure be executed in parallel. This in turn requires that the procedure be divided into several parallel threads executed concurrently, then brought back together to compute the final result.
peak workflow throughput
In the Capacity Planning Tool, this term refers to the number of work units completed during the estimated peak workflow period. Peak workflow throughput is calculated by multiplying the identified peak users by the set peak client displays per minute, and the result is identified in terms of peak displays per minute for each workflow (peak users x DPM/client = DPM). The capacity planning models used in the CPT are based on the peak workflow throughput values.
A general term referring to the efficiency of the computer or system to do work. Performance components include processing time (service times and network transport times), queue time (random arrival wait time), travel time (latency), and think time (time for user interaction with the display). Display response time includes the processing time, queue time, and any additional travel time.
A specific reference platform performance represented by a SPEC performance benchmark value. Performance baseline values for workflow service times used in the ESRI Capacity Planning Tool are established each year based on the performance of available vendor platform technology. ESRI performance baselines are identified by the term “Arc” followed by the year (i.e., Arc13, Arc14)
A test run of a computer program that takes the measure of the execution performance of that program. Both the test and the measurement of the test are called a performance benchmark. Performance benchmarks are used to identify hardware processing loads generated by the software when executing a defined procedure. Platform vendors publish performance benchmarks to represent the relative processing capabilities of their hardware technology. Performance benchmarks are also useful for identifying changes in software performance from one release to the next, and can be used as a reference for establishing capacity planning performance targets.
The workflow software component service times used to establish a projected performance and scalability baseline. Component software performance should be reviewed during development, prototyping, and deployment to verify that application performance is within the established performance budgets.
A representation of the relationships between the various system performance factors, which can be used for platform sizing and predicting appropriate network bandwidth. In this book, performance models are in the form of a chart, a table, or an Excel interface.
The workflow software response times used to establish a projected performance and scalability baseline. Component software performance should be reviewed during development, prototyping, and deployment to verify that application performance is within the established performance targets.
The hardware platform and software component configuration supporting the production software environment. The platform architecture also defines the protocols used to communicate between the software components and how they are configured.
Identifies the peak platform throughput. The Capacity Planning Tool identifies peak platform throughput values based on the selected workflow. In a system environment where multiple workflows are supported on a single platform, the peak platform capacity value is projected based on maintaining the same relative workflow throughput mix.
The term used to identify and select a specific platform component configuration. The Capacity Planning Tool identifies the chip manufacture (Intel), the processor model number (5160), the number of processor cores (4 core), the number of chips (2 chip), and the processor MHz and L2 cache [3000(4) MHz]. Platform memory requirements are generated by the CPT using standard guidelines based on system capacity requirements. The platform configuration naming convention is used to identify a specific processor configuration, which may be provided by more than one hardware vendor (i.e., Intel 5160 4 core (2 chip) 3000(4) MHz) platform configuration is provided by Dell, IBM, HP, and Sun).
The total processing time required to support the identified user requirements workflow profile. A composite baseline display service time is computed based on the identified user requirements workflow and allocated to each appropriate platform tier. The computed baseline platform display service time is adjusted for the selected platform performance using the vendor published SPEC benchmark values. The adjusted platform service time is used to calculate resulting platform performance values.
The term used to identify the number of platform servers included in the platform configuration. The CPT will generate the required number of platform nodes for each platform tier based on the identified user workflow requirements and the selected platform configuration. If a fixed number of nodes is identified, the platform performance will be calculated based on the fixed node value.
platform operating system
The platform operating system (OS) is the software that manages the sharing of the resources of a computer and provides software programs with an interface used to access those resources. An operating system processes system data and user input, and responds by allocating and managing tasks and internal system resources as a service to users and programs of the system.
platform service times
See also service time. Refers to the platform processing load required to support the selected workflow configuration. The workflow software component processing loads (service times or the platform performance baseline) are allocated to the assigned platforms to establish the baseline platform service times. The baseline platform service time is translated to the selected platform service time using the SPEC relative benchmark values. The selected platform service time and the user requirements peak workflow are used to calculate the platform performance values.
A process informed by the individual platform sizing charts provided in chapter 8 (Platform performance). These charts identify individual platform capacity based on the standard ESRI software workflow models and platform SPEC performance benchmarks provided by vendors. Platform sizing identifies platform capacity requirements and is useful for comparing several platform alternatives on a single display. These charts use the same capacity planning models incorporated in the CPT, and were used for capacity planning by ESRI system design consultants since the mid-1990s.
platform sizing chart
The platform sizing chart can be used to identify peak platform throughput based on vendor provided platform SPEC performance benchmarks. Platform sizing charts are provided in chapter 8 for the ArcGIS Desktop Windows Terminal Server, ArcSDE Geodatabase Server, and Web Services (ArcIMS, ArcGIS Server) platform environments. The platform sizing charts do not provide the workflow response times, software component service times, queue times, network transport times, network latency, or network traffic requirements that are included in the Capacity Planning Tool.
A group of common server platforms configured for a scale-out platform architecture. Web services are deployed in a two-tier (Web server, database server) or three-tier (Web server, container machine, database server) architecture. A single-tier workgroup server tier is provided for a single platform configuration hosting the Web applications and database on the same platform. Windows Terminal Server farms host centralized data center deployment of Windows desktop applications, and are identified as their own platform tier.
The hardware vendor that builds and sells the supported host server platforms. Platform vendors also conduct and publish the SPEC performance benchmark results for their own hardware. The vendor-published SPEC benchmark values are used to identify relative platform performance in the Capacity Planning Tool.
process queue time
The waiting in line for processing time is called queue time. Process queue time is used to identify the total system wait times resulting from the random arrival distribution of individual process transaction requests. When multiple requests arrive for processing by a single service provider (core processor), only one request can be processed at a time and the others must wait in the queue.
The work done by the computer in executing the software program instructions. The core processor unit is the computer hardware component that executes the software instructions. The amount of time the core processor is busy executing software instructions is called the processing time, or service time spent by the computer to process the program instructions.
cache A cache used by the central processing unit of a computer to reduce the average time taken to access memory. The cache is a smaller, faster memory, which stores copies of the data from the most frequently used main memory locations. Hardware vendors design computer chips to provide the optimum performance and capacity by reducing the chip size and optimizing the chip configuration.
The computer processor component used to execute the software program instructions. The core implements optimizations such as superscalar execution, pipelining, and multithreading. Multicore chips provide a single integrated circuit that can include two, four, or more cores on a single integrated silicon chip configuration.
The speed (clock rate) at which a microprocessor executes instructions. Computer chips include an internal clock that regulates the rate at which instructions are executed and synchronizes all the various computer components. The core processor design optimizes the number of program instructions that can be executed during a clock cycle. For a specific processor core design, faster clock speeds provide faster program execution. Clock speeds are expressed in megahertz (MHz) or gigahertz (GHz).
The connector on the computer motherboard that houses a computer processor. Processor sockets use a pin grid array (PGA) where pins on the underside of the processor connect to holes in the processor socket.
See user productivity.
The mathematical study of waiting lines (or queues). Queuing theory is an academic study of distribution arrival times, and provides a definition of random arrival distributions used in supporting a variety of business decisions. Statistical arrival distributions are difficult to quantify. Mathematical queuing models are most reliable when estimating large, random-population arrival distributions.
The system wait times resulting from the random arrival distribution of individual process transaction requests. When multiple requests arrive for processing by a single service provider (core processor or shared network port), only one request can be processed at a time and the others must wait in the queue. The waiting in line for processing time is called queue time, and can be an indication of overcapacity conditions.
random arrival times
An arrival distribution where all available arrival times are equally likely. Queuing theory can be used to estimate wait times for a random distribution of arrival times. In computer processing, hundreds of communications are transmitted to support a single user display transaction. These display transactions typically occur over a period of less than 5 seconds. For systems supporting many concurrent user workflows, it is quite probable that the resulting transaction arrival times would follow a random distribution profile. Queuing theory can be used to estimate the probability that multiple transaction requests will arrive for processing at the same time, and will have to wait in line for processing. Queuing theory can be used to estimate the wait time (queue time) for capacity planning purposes. An approximation of the queue time (a function that depends on component utilization rates) is included in the CPT to account for these random arrival times.
The term used for identifying REST service path names that are published with ArcGIS Online or Portal for ArcGIS content. Registered services can be published by a named user with publishing or contributing privileges. A registered service client request to ArcGIS Online or Portal for ArcGIS would receive the registered service transaction from the original service provider.
The term used in comparing the productivity of two common service providers. The SPEC performance benchmarks were established to provide a fair measure of relative compute performance of one hardware vendor platform configuration compared to another, for capacity planning purposes. A selected platform is identified to represent the performance baseline, and the SPEC published benchmark value represents the performance of the test platform relative to the baseline platform.
remote desktop services
Introduced by Microsoft to replace the term Windows Terminal Server.
User locations that are located over a remote network connection (WAN or Internet). A local site is a user location that is located on the same LAN environment.
The process for identifying peak user workflow requirements. The Capacity Planning Tool includes a requirements analysis module that can be used as a template to identify peak user workflow requirements. User workflow software component service times are identified on the CPT workflow tab, and then used to complete the user analysis within the CPT user requirements module.
The total time from initial transaction request to final delivery of the service. For a user map display, this includes all system processing times (service times), data transfer times, queue times, and any transport latency times.
Refers to software programs that include a single threaded execution (execute the software procedure one step at a time). Each step (instruction) in the procedure (program) is executed in sequence, and each instruction in the sequence must be completed before moving on to the next instruction.
A hardware platform used to host a software program that performs a compute service. A server is normally a shared platform environment that hosts multiple concurrent user workflows during a common service period.
Any computer system component (server platforms or communication networks) that process workflow service requests (user displays or program transactions). (The workflow units used in the CPT are called user displays.)
See also platform service times. The transaction processing time (usually identified in seconds) required to service a single workflow unit of work (called a user display or workflow display). A performance baseline workflow display can be identified as the work of a power user performing a workflow load over a period of six seconds (10 displays per minute).
Refers to the fact that a single processor core is limited to executing a single batch process thread during a single clock cycle. Multiple concurrent processing requests must wait in a process queue and will be executed sequentially by a single processor core. A server platform configuration with 4-core processors can execute up to four concurrent process requests at the same time, each core limited to executing a single process execution.
The bandwidth of the site router or gateway connection to the WAN or Internet communication service provider. The cost of the connection service is usually based on the amount of connection bandwidth provided by the service.
service times Refers to the performance baseline processing times used to represent the software workflow model. Standard ESRI workflow software service times are provided on the CPT Workflow tab based on performance capabilities of the core GIS software. The ESRI workflow service times can be used as a template to establish specific project workflow performance budgets. The software baseline service times identified for the project workflows establish the performance models used in the CPT system loads analysis. These workflow baseline service times represent the software component performance budget represented by the CPT capacity planning analysis.
software service times
Refers to both the software baseline service times used in establishing the workflow performance model and the adjusted software service times (corrected for selected platform performance) represented in the Workflow Performance Summary chart included on the CPT2006 tab.
SPEC baseline performance value
The reference SPECrate_int2006 platform performance baseline used for publishing the standard ESRI workflow software service times and all other customized workflows included on the CPT Workflow tab. The SPEC baseline performance value is updated each year to represent the current platform hardware performance published on the SPECint_rate2006 benchmarks. SRint2000 A shorthand notation for the published vendor SPECrate_int2000 platform benchmarks. The SRint2000 benchmarks were used to represent vendor platform relative performance from year 2000 through December 2006. Hardware platform performance comparisons after 2006 use the SRint2006 benchmarks.
A shorthand notation for the published vendor SPECrate_int2006 platform benchmarks. The SRint2006 baseline benchmark values are used for relative platform performance service time adjustments in the Capacity Planning Tool. SPEC performance values for vendor platforms sold from 2000 through 2006 would be found in the SRint2000 benchmark results. The SRint2006 results can be estimated from the SRint2000 results by dividing by a factor of 2.1, allowing approximate performance comparisons between the older and newer platform environments.
standard ESRI workflows
The term used to identify the software technology workflow templates included on the CPT Workflow tab. These workflows can be used as a reference for establishing project-specific workflows during a capacity planning analysis. Workflows with increased software service times represent more conservative performance targets, and workflows with reduced software service times represent more aggressive performance targets. The standard ESRI workflow software service times represent reasonable performance targets for the associated ESRI software technology.
A term used to describe a persistent storage media (normally on disk or tape) that retains digitally encoded data. Current platform storage solutions are provided by hard disk devices that provide a persistent storage repository for digitally encoded data. Storage can also include computer tape, compact disk (CD), and memory, which also provide temporary or persistent media storage environments. The term storage used in capacity planning normally refers to either internal server storage, direct attach storage, storage area networks, and network attached storage environments supported by a redundant array of independent disk (RAID) hardware configurations.
A term used to describe business systems that support their data on internal disk drives rather than sharing their data on a storage area network (SAN). Hardware funds are invested to support a single project effort, rather than shared to support more optimum corporate-level business needs. Stovepipe systems do not connect to the rest of the community, and it is difficult to share business resources with this type of architecture.
See Windows Terminal Server.
The term used to describe the rate of doing work, expressed as a transaction rate, such as transactions per minute. In communication networks, throughput is the amount of digital data per time unit that is delivered over a physical or logical link, or that is passing through a certain network node. For example, it may be the amount of data that is delivered to a certain network terminal or host computer, or between two specific computers. The throughput is usually measured in bits per second (bits/s or bps), occasionally in data packets per second or data packets per timeslot. The system throughput or aggregate throughput is the sum of the data rates of delivery to all terminals in a network. The maximum throughput of a node or communication link is synonymous to its capacity.
In computer platforms, throughput is the number of displays per minute (DPM) hosted by the computer platform tier or supported on a certain platform node. For example, throughput could be the amount of processing required by a host computer or by a server platform tier. The throughput is measured in displays per minute (DPM), wherein each display represents a single unit processing load specific for each defined user workflow. The system throughput or aggregate platform throughput is the sum of the workflow processing times, where the platform display (unit of work) is a weighted average of the total platform processing load. The maximum throughput of a node or server tier is synonymous to its capacity.
See also bandwidth, capacity, network throughput, network traffic, peak workflow throughput, platform capacity, platform sizing chart, workflow capacity.
The term used to describe a unit of work (for GIS workflows, a transaction may be represented by a map display). Each user workflow may have a different measure of work. The transaction provides a unit for measuring throughput, service time, queue time, and response time—all measurements are relative to the type of workflow transaction.
The term used to represent the displays per minute (DPM) performed by the average user for each specific workflow included in the CPT requirements analysis. User productivity for each workflow is defined on the CPT Workflow tab. Recommended maximum user productivity for ArcGIS Desktop workflows is 10 DPM, while recommended maximum user productivity for Web application workflows is 6 DPM. User productivity is restricted by workflow display response time and required user think time. Minimum user think time must always be positive, and should provide the user with sufficient user interaction time to review the display and provide appropriate user input. User productivity should be adjusted as needed to accommodate a realistic user think time.
user think time
The human interaction time included in the user workflow to accommodate the required user input for each display cycle. Available user think time is calculated by first computing the display cycle time (60 seconds / user productivity) and then subtracting the display response time (total time required for the system to return the user display). For user workflows, the user think time must always be greater than zero. A minimum think time of 3 seconds is used as a default for standard workflow settings on the CPT Workflow tab. Minimum time can be adjusted to represent an average minimum user think time for each workflow, and is used to support the conditional color warnings in the user productivity column (red when less than minimum user think time). A workflow with zero think time is a batch process (no user interaction time).
Used here to identify the configuration of the GIS software technology employed to enable the user work activity. A separate user workflow can be identified for each use-case or information product description involved in quantifying user technology requirements in the user needs assessment. Some workflows may be employed to generate a broad range of information products using the same software technology and data source. A single unit of work is identified as a workflow display, and is quantified by specific baseline component software processing time and the generated network traffic per display (unit of work). For capacity planning purposes, a desktop software workflow display transaction is defined in terms of component software service times (desktop, WTS, SDE, DBMS) and network traffic (Mbpd) generated by a power user during an average period of 6 seconds (10 DPM productivity). A Web software workflow display transaction is defined in terms of component software service times (WA, SOC, SDE, DBMS) and network traffic (Mbpd) generated by a power user during an average period of 10 seconds (6 DPM productivity).
The term used for identifying core ArcGIS Online or Portal for ArcGIS services. Typical utility services include Printing, Geocoding, Geometry, and Route services.
For hardware platforms, the term used to identify the percentage of time the platform core processors are busy executing the software program instructions. The processing time (service time) is computed by comparing the time the core processors are busy over the total work unit cycle time (i.e., 10 percent utilization of a single core processor over a 6-second work cycle time would represent 0.6 seconds service time). Platform utilization approaches 100 percent as platform throughput reaches full output capacity. (Another way of putting it is that utilization is a measure of platform capacity consumed by a specific throughput, and is represented as a percentage of platform capacity. Peak platform throughput is achieved at 100 percent utilization.)
For network communications, the term is used to identify the percentage of time the network connection is busy while supporting the required network traffic. Network utilization approaches 100 percent as network traffic throughput reaches bandwidth capacity.
Web application server (WA)
The server platform hosting the Web application software when configured in a three-tier Web architecture. The Web application server (WA) is identified with the platform service times in the User Requirements Module, and the Web application server tier is included as one of the platforms in the Platform Configuration Module. The Web application platform service times are included with the Web spatial server (SS), and the Web application software is hosted on the container machine platform tier in a two-tier Web architecture.
A client software application that enables a user to display and interact with text, images, maps, videos, and other information typically generated by a Web application over the World Wide Web or a local network.
A server platform that hosts Web applications.
wide area network (WAN)
A computer network that covers a broad area; any network whose communication links cross metropolitan, regional, or national boundaries. Or, less formally, a network that uses routers and public communications links. Contrast with personal area networks (PANs), local area networks (LANs), campus area networks (CANs), or metropolitan area networks (MANs), which are usually limited to a room, building, campus, or specific metropolitan area (e.g., a city), respectively. The largest and most well-known example of a WAN is the Internet.
Windows Terminal Server (WTS)
The term includes a reference to Microsoft Terminal Services and the Citrix XenApp (previous Citrix Presentation Server) technology. Terminal Services is a component of Microsoft Windows (both server and client versions) that allows a user to access applications and data on a remote computer over any type of network. The Microsoft Remote Desktop Protocol, part of Microsoft's Terminal Services, is based on Citrix technology and was licensed from Citrix in 1997. The Citrix XenApp is built on the Independent Computing Architecture (ICA), Citrix Systems' thin client protocol. Unlike traditional frame-buffered protocols like VNC, ICA transmits high-level window display information, much like the X11 protocol, as opposed to purely graphical information. This is possible because the Citrix Display Driver, which is installed in Session Space, is capable of capturing high-level GDI draw commands that can be replayed on GDI-capable clients (for example, Windows-based clients). Clients are available for several operating systems, including Microsoft Windows (CE, 16-bit, 32-bit and 64-bit platforms), Mac OS, Linux, and other Unix-like systems.
An artificial term used by the Capacity Planning Tool as a measure of an average unit of work. The CPT defines each unique user workflow by identifying the computer processing time (software service times) and data traffic for an average work transaction. For simple GIS map displays, the work transaction may be identified as the work (computer processing and data transport) associated with the average map display. For more complex workflows, the work transaction may be identified as the work (computer processing and data transport) of a power user over a specified period of time (6 seconds). The work transactions for each user workflow are located in the software service times section of the CPT Workflow tab.
The peak workflow throughput of a selected server platform configuration.
Any of the performance values (service time, queue time, response time, etc) associated with a specific workflow on a selected server platform configuration.
workflow performance summary
An Excel bar chart included in the CPT User Requirements Module that identifies workflow service times, queue times, network transport times, network latency, and total response time based on identified user requirements, network bandwidth, and selected platform architecture.
Information required by the CPT User Requirements Module to identify the peak user workflow loads. This includes the definition of each workflow on the CPT Workflow tab, selection and location of user workflows on the CPT2006 Capacity Planning tab, and identification of peak users and productivity. The CPT can be used as a template for identifying user workflow requirements.
workflow service times
The work transaction software service times identified on the Workflow tab that define a user workflow. The CPT applies the workflow service times to the identified user requirements to complete a system loads analysis. The system loads analysis identifies peak network traffic and processing loads on the selected platform configuration.
The client desktop computer.