Glossary

A

Add My Capacity plan

To add the free capacity of your own computers onto your Parabon rate schedule.

average demand

The demand as determined by dividing the total number of cap-hours by the number of units of time in the interval.

B

Benchmark

A fixed amount of computational work used to measure the performance of an engine or a computer running it. See Parabon Standard Benchmark.

billing demand

The demand upon which billing is based, as specified in a rate schedule or contract. It may be based on the contract year or a contract minimum, and therefore does not necessarily coincide with the actual measured demand of the billing period.

billing period

The interval of time between invoices.

C

cap (C)

The cap (constant applied power - symbol: C) is the Parabon unit of applied computational power (or potential computational capacity), equal to the rate of computational work performed by a computer with a Parabon Capacity Index of 1.0.

For more information, please see the Pricing FAQ.

capacity

See computational capacity.

capacity index

See Parabon Capacity Index (PCI).

cap-hour (Ch)

The amount of computational work that can be performed in one hour by an average system — i.e., a node that provides 1 cap (C) of computational power.

NOTE: Above-average machines (such as those with faster processors or a greater number of CPUs) often complete 1 Ch of work in less than 60 wall-clock minutes, and/or perform more than 60 cap-minutes of work in a single hour.

For more information, please see the Pricing FAQ.

checkpoint

A snapshot of a task that can be used to resume is at the point it is interrupted, so it can be resumed later from that exact position.

client

Contrary to the nomenclature used by some in the grid computing community, at Parabon a client is a user-side application that interacts with a Frontier grid service. The term client is not a synonym for a Frontier Compute Engine because compute engines, and the computers on which they are hosted, are not clients of the service; rather, they provide capacity to the service.

client application

A program that uses Frontier via the Frontier Client API.

cloud services

Dynamically scalable, long-running applications hosted on virtualized infrastructure. Examples include web and business application or transaction processing services.

complementary time zones

The circadian rhythm of human life results in waves of computational demand each day, peaking in time zones where people are awake. For a given time zone T, the set of time zones {~T} that are complementary to T are defined as those in which people are asleep while people in T are awake. Naturally, demand in T is most easily satisfied by capacity from {~T}, which is why Parabon contracts with providers from around the globe.

computation

See computational work.

Computation on Demand® service

Parabon's computation on demand utility service.

computational capacity

The computational power a computer can apply to execute work.

computational demand

The rate at which a job uses, a customer consumes or a grid delivers power over any designated period of time.

computational grid

A computational grid aggregates capacity from multiple networked resources and delivers it in the form of a utility service. Most computational utilities are provided from a data center. The Parabon Computation Grid is the only computation utility powered by excess capacity that is purchased from and supplied by individual and institutional providers from around the globe. This allows for far greater scalability and, because excess capacity is abundant, it is the most affordable source for high-performance computation.

computational power

The rate at which computational work is performed per unit time.

computational work

A finite series of software instructions performed by a computer. It is also the application of computational power over time. Hence, work = power x time.

compute engine

See Frontier Compute Engine.

compute-to-data ratio

The compute-to-data ratio for a task refers to the amount of time required for its uninterrupted execution on an engine versus the time required for its network transmission (both the time for data and code to be transmitted to the engine and the time required for the return of final results). The compute-to-data ratio for a job is simply the average of the compute-to-data ratios of its tasks. Applications ideally suited to Frontier are those with high compute-to-data ratios.

CPU-hour

A measure of computer power meaningful only in the context of a set of homogenous computers.

Crush™

See Parabon Crush™.

customer maximum 5-minute demand (peak demand)

The greatest rate at which a customer has consumed power during any period of 5 consecutive minutes in the current billing period.

D

demand

See computational demand.

demand charge

That part of the charge for the computation service that is based on a customer's demand.

demand interval

A period of time during which power is measured for determining demand.

demand rates

A charge for computation that is based upon a maximum demand during a billing period.

Distributed Network Attack®

AccessData's scalable enterprise solution, Distributed Network Attack® (DNA®), enables organizations to employ computers across their IT infrastructure to decrypt over 70 different file types. See Distributed Network Attack Enterprise.

E

engine availability requirements

The requirement for a provider to make engines under contract to Parabon available for task execution during specified time periods.

Evolutionary Computation (EC)

An optimization technique that employs genetic algorithms, for example, to iteratively generate improved result sets, based on a given fitness function.

Parabon's Origin™ Evolutionary SDK is a framework for quickly developing EC-based applications.

excess capacity

The unused power of dormant computers that is harnessed by the Frontier Compute Engine.

F

Flexible Capacity Schedule

Parabon's pay-per-use rate schedule.

FLOPS

Floating Point Operations Per Second. A common computational performance metric. Click here for a discussion of FLOPS versus the Parabon Capacity Index.

Frontier API

Provides the framework for creating, launching, monitoring, and controlling compute-intensive jobs from an average desktop computer. Consisting of two primary components - the Task Runtime API and the Client API - the Frontier API enables developers to quickly build new applications or adapt existing applications to run on Frontier. Click here for more information on these APIs.

Frontier Client API

The interface between a client application and Frontier allowing the creation, monitoring, and control of jobs and tasks.

Frontier Compute Engine

Parabon's desktop application that unobtrusively harnesses the excess capacity of a host computer for use on a Frontier Grid. It retrieves work from a Frontier server, manages its execution on its host and returns results to the server upon completion.

Frontier Enterprise

The intranet version of Frontier.

Frontier Grid

A collection of nodes together with a Frontier Server that can perform jobs. A Frontier Grid may be a Public Grid, Private Grid, or Virtual Private Grid.

Frontier Grid Platform

Parabon's extreme scale grid computing platform that harnesses the excess computational capacity of networked computers (desktops, servers, and even mainframes) and aggregates it to deliver supercomputing capacity.

Frontier Grid Services

The various grid services supported by Frontier. See Frontier API for more information.

Frontier IDE for Eclipse

Parabon's Integrated Development Environment (IDE) packaged as an Eclipse plug-in and used for developing Frontier-powered applications.

Frontier Rapids™

Frontier Rapids™ is an integration and execution environment for rapidly deploying existing applications across Frontier. Simply drop your binary or jar file into a Rapids Runtime directory, customize a Rapids Configuration File and experience the power of Frontier faster than you thought possible. Many existing applications can run on Rapids with no code changes! And, its job management commands provide your application with the range of control over Frontier that you'd expect from a fully integrated Frontier application. Whether you want a quick proof-of-concept or you need grid-scale capability yesterday, race onto the Frontier Grid Platform with Frontier Rapids™.

Frontier SDK

Parabon's Frontier API and other software tools used for the development of Frontier-powered applications. Click here to learn more.

Frontier server

One or more computers running the Frontier server software.

Frontier server software

Server software that communicates with Frontier-powered applications and Frontier Compute Engines to manage the execution of jobs on a Frontier Grid.

G

grid

See computational grid.

grid load

The aggregate demand on a grid at a point in time or over a designated period of time.

I

inSēquio™

Parabon's CAD (computer aided design) application that optimizes synthetic DNA sequences for creating self-assembling nanostructures. Click here for more information.

institutional providers

A registered provider that supplies capacity (50 C minimum) to Parabon in return for compensation.

Integration and Execution Environment

The Frontier Rapids™ IEE allows existing applications to be quickly and easily deployed across Frontier.

intermediated computation

Computation supplied by an intermediary, like Parabon, that aggregates capacity from multiple providers.

IOS

The inSēquio Optimization Specification (IOS) is an XML file specification used by Parabon's inSēquio™ Design Studio for storing nanostructure designs and optimization criteria.

J

job

A batch of work that, if it is to benefit from execution on a computational grid, must be decomposable into independent units of work called tasks. Examples of grid jobs include modeling & simulation, data mining, and predictive analysis, etc.

job priority

The part of the charge paid by a client for the aggregate number of cap-hours applied to jobs.

job results

The results/output of a job.

JobController

A utility program that comes as part of Frontier SDK that may be used to monitor jobs and individual tasks running on Frontier.

K

kilocap (kC)

Equal to 1,000 caps.

kilocap-hour

The amount of work that can be performed in one hour with 1 kC of power.

M

maximum demand

The greatest demand during a specified period of time such as a billing period.

MIPS

Millions of [integer] instructions per second. Click here for a discussion of MIPS versus the Parabon Capacity Index.

monthly reserve payment

A component of the monthly compensation paid to provider for an engine that has met its availability requirements.

monthly service charge

A recurring, non-refundable fee charged on a monthly basis to help cover the fixed costs associated with maintaining an account.

N

nanoscale

One thousand times smaller than microscale. Literally, the scale of individual atoms and molecules.

native sandbox

Virtual machine technology used by the Frontier Compute Engine to securely execute native codes (e.g., C/C++, Fortran).

node

A computer running the Frontier Compute Engine software for the purpose of performing computational work toward the completion of jobs.

O

Opportunistic Evolution

Opportunistic Evolution (OE) is a novel evolutionary algorithm, available in the Origin™ Evolutionary SDK, designed to maximally utilize a given set of grid resources. It is used in several of Parabon's applications.

P

Parabon Capacity Index (PCI)

A unitless scalar representing an engine's performance rating normalized by the mean performance rating of all active engines.

Parabon Computation Grid

Parabon's large Internet-based computation grid that powers its Frontier Grid Services.

Parabon Crush™

Crush™ is a large-scale data mining application for deep statistical modeling and knowledge discovery.

Parabon Origin™ Evolutionary SDK

A Java-based SDK for quickly developing evolutionary computation (EC) applications on Frontier. Click here to learn more about Origin.

Parabon Standard Benchmark

The task used by Parabon to benchmark the computational power of engines or the computers on which they run.

payment period

The interval of time between payments made by Parabon to a provider.

performance rating

The inverse of the time required for an engine to execute the Parabon Standard Benchmark.

power

See computational power.

provider

Individuals or entities that provide excess capacity to Parabon, via the Frontier Compute Engine, from one or more computers over which it has full usage authorization.

R

reserved capacity rate schedule

The Parabon rate schedule that applies when users reserve specific levels of capacity for specific time periods.

T

task

A portion of a job suited for independent execution on an engine.

task attributes

A task's identifying information including when it was created.

transmission charge

The charge to transmit job information across the Parabon Computation Grid.

transmitted job information

The aggregate number of bytes of information (code, libraries or data) transmitted as part of a job along any of the four communication channels of a Frontier Grid (User-to-Server, Server-to-Compute-Engine, Compute-Engine-to-Server and Server-to-User).

U

usage

The amount of computational work, as measured in cap-hours, performed by a engine during a given period of time. Importantly, usage does not refer to wall-clock time. In 30-minutes of continuous processing the usage of a engine with a PCI of 2.0 is 1.0 cap-hours.

usage charge

The part of the charge ascribed to the aggregate number of cap-hours applied to a users jobs during a billing period.

user

A person who executes applications on a Frontier Grid.

V

Virtual Appliance

A self-contained virtual machine (VM) image bundle, containing everything necessary to instantiate a new virtualized guest OS in either a VMware vSphere or Citrix XenServer cloud infrastructure.

W

Watchman™

Watchman™ uses Evolutionary Computation techniques to optimize the placement of sensors for coverage over a user-specified area of interest based on building and terrain data. See Watchman.

work

See computational work.

Learn More about Computation on Demand®

Want to know about anything else?

Contact us. We're eager to help.