Panel on:

DEFINING the “GRID”

Moderator: Frederica Darema, NSFPanelists:Fabrizio Gagliardi, MicrosoftIan Foster, ANL & UofChicagoGeoffrey Fox, Indiana U.Filia Makedon, NSF & DartmouthEd Seidel, Louisiana State U.

2

Intent of this discussion

• This panel is intended as a venue to present some views and open a dialogue for discussion

• What is a “grid”?– what’s the fuss about it? – scope? several views of a grid? – past, present, emerging grids?

• Define a process for a comprehensive, concise definition of what’s a grid

3

What is Grid Computing?

coordinated problem solving on dynamic and heterogeneous resource

assemblies

QuickTime™ and a decompressor

are needed to see this picture.

QuickTime™ and a decompressor

are needed to see this picture.

IMAGING INSTRUMENTS

COMPUTATIONALRESOURCES

LARGE-SCALE DATABASES

DATA ACQUISITION ,ANALYSIS

ADVANCEDVISUALIZATION

Example: “Telescience Grid”, Courtesy of Ellisman & Berman /UCSD&NPACI

4

A bit of history…and trends…

• The “grid concept” which in the mid-90’s was the purview of a few academics is now entrenching the industrial sector

• In some ways the grid started systematizing ideas that preceded it– like distributed computing environments (DCEs)

• The first demos of “grids” starting with the I-Way at SC’95

• Followed by several demos (~97-…)of distributing a single application on multiple, geographically distributed nodes

5

In the (recent) beginning..• The efforts that started in the mid-90’s had an emphasis

more akin to the parallel, cluster, NOWs, and metacomputing environments. – they were a step beyond the high-end (supercomputer)

parallel, or cluster, or NOW systems – placed emphasis on creating computational capabilities

where a single application can be partitioned and deploy multiple, heterogeneous and geographically distributed platforms,

– where each of these “nodes” could in itself be a high-end (supercomputer) system,

• in other words, the grid was originally viewed as a means of increasing the computational (mips & flops) power, – beyond the high-end or supercomputers of that time.

• The dominant term for such grids has been: computational grids.

6

Revisiting and building on the past…

• Together with efforts that built upon the notions of DCEs, and as computational grids’ environments enabled larger applications to execute, few other aspects have become apparent:1) larger capabilities also entailed larger and often distributed sets of data (consumed or produced by such applications) and 2) application models that incorporate multiple modalities of the application system (complex application models).

• These aspects, pointed to other modalities of “grids”, more akin to the notions of DCE’s, where grids enable executing the various models of complex applications on different (and perhaps specialized) platforms of such a grid.

• Notions like workflow, which were popular in the late 80’s and early 90’s for CIM (Computer Integrated Manufacturing) applications are again finding their way to the present grids.

• Notions of VMs also are brought again in the forefront• Furthermore grids are becoming coupled with the notion of

collaboratories, leading to notions of various grid communities….

7

Moving forward…

Dynamic Data Driven Application Systems(DDDAS)

A new paradigm for applications/simulations and measurement methodologies

PROGRAM SOLICITATION NSF05-570 Sponsored by NSF, NIH, NOAA

Cooperating Programs: EU Grids & e-Infrastructure, and UK e-Sciences

Solicitation announced on March 10, 2005Proposals received: June 13, 2005Awards made by end of Sept, 2005

8

Measurements ExperimentsField-Data

User

Theory

(First P

rincip

les)

Simula

tions

(Math

.Modeli

ng

Phenomenol

ogy)Experiment

MeasurementsField-Data

(on-line/archival)User

Theory

(First P

rincip

les)

Simula

tions

(Math

.Modelin

g

Phenomenolo

gy

Observ

ation M

odeling

Design)

OLD

(serialized and static)

NEW PARADIGM

(Dynamic Data-Driven Simulation Systems)

Challenges:Application Simulations DevelopmentAlgorithms Measurement Instruments InterfacesComputing Systems Support

Dynam

ic

Feed

back

& C

ontro

l

Loop

What is DDDAS(Symbiotic Measurement&Simulation Systems)

9

DDDAS: Beyond Grid Computing New Capabilities in Applications and

Measurements--dynamic integration between applications&measurements---

QuickTime™ and a decompressor

are needed to see this picture.

QuickTime™ and a decompressor

are needed to see this picture.

COMPUTATIONALRESOURCES

DATA ACQUISITION ,ANALYSIS

ADVANCEDVISUALIZATION

10

Beyond Grid Computing “Extended Grid’:

the Application Platform is

the computational&measurement system

Applications

Com

puta

tion

al

Plat

form

s

Inst

rum

ents

Sens

ors

Archi

val/

Stor

ed D

ata

Measurements Computational Grids

11

Examples of Programmatic Support:• Programs like

– the DARPA QUORUM Program (1996-2002) were motivated to support grids along the directions of DCE, RT, &… also supported computational grids.

• While programs like– the DARPA Systems Environments (1996-1998), – the NASA Information Powergrid (IPG, 1996-2002), – and later DOE Science Grid - SciDAC program (2000 – present?) placed emphasis in enhanced computational capabilities provided by grids,

• More recently:– the NSF Next Generation Software Program (NGS, 1998-2004) – the NGS follow-up Computer Systems Research (AES&SMA) (2004-present) – and the NSF Middleware Initiative (2000-present) emphasize both aspects of the grids, the first two with research emphasis the

later with infrastructure emphasis. • Emerging and future directions, Dynamic Data Driven Applications

Systems (DDDAS), (2000 -…) further extend the present notions of the “grid”

• PLUS several Programs in EU, UK, Asia, AU

12

The challenge and the opportunity

• All these aspects and grid modalities give the impression of a non-coherent view of what is the “grid”

• it looks like the proverbial ‘parts of an elephant’. • Furthermore, emerging technologies and concepts, such as:

– sensors and sensor networks, – dynamically integrated computational and observational

systems

point to environments which further extend the grid notions• Do we have multiple grid instantiations? an “amoeba grid”?• But even so can we have a comprehensive all encompassing

definition?• For these reasons, it behooves us to have a discussion for

defining a concerted notion of the “grid concept”

13

• This panel, consisting of developers and users of grid technologies advances, and notable visionaries, will discuss their views of what’s the “grid” and how to lay the foundation of a definition for the grid

14

15

ExperimentMeasurements

Field-DataUser

Simula

tions

(Math

.Modelin

g

Phenomenolo

gy

Observ

ation M

odeling

Design)

Dynam

ic

Feedbac

k

& C

ontrol

Loop

DDDAS has potential

for significant impact in

science, engineering, and commercial world,

akin to the transformation effected

since the ‘50s

by the advent of computers

NSF05-570www.cise.nsf.gov/dddas

Proposal Deadline: June 13, 2005

Enabling the DDDAS vision rests on

the creativity and resourcefulness of the research community