Grid Computing

GRID COMPUTING

PROFILE

Name

:

Name of the Institution

: DEPARTMENT OF COMPUTER SCIENCE

1) ALEKYA REDDY 2)G NITISH KUMAR (3/4B.Tech CSE)

ELLENKI COLLEGE OF ENGG. AND TECH.

Title of the Paper E - Mail

:

:

GRID COMPUTING

[email protected]

1

CONTENTS PG NO ABSTRACT

4

1. INTRODUCTION

5

2. THE GRID

5

3. BEGINNING OF GRID

5

4. GRID COMPUTING CHARACTERSTICS

6

5. A COMMUNITY GRID MODEL

7

6. TYPES OF GRID

8

7. THE KIND OF GRID TOOLS

8

8. GRID COMPONENTS

9

9. ADVANTAGES

14

10. CHALLENGES OF GRID

14

11. CONCLUSION

14

12. BIBLIOGRAPHY

15

2

LIST OF FIGURES

FIG NO

NAME

PG NO

1

Layered architecture of grid model

7

2

Possible user view of Grid

10

3

Security in Grid Environment

10

4

Broker Service

11

5

Scheduler

12

6

Data Management

13

7

Gram

13

3

Abstract

Grid computing, emerging as a new paradigm for next-generation computing, enables the sharing, selection, and aggregation of geographically distributed heterogeneous resources for solving large-scale problems in science, engineering, and commerce. The resources in the Grid are heterogeneous and geographically distributed. Availability, usage and cost policies vary depending on the particular user, time, priorities and goals. It enables the regulation of supply and demand for resources. It provides an incentive for resource owners to participate in the Grid; and motivates the users to trade-off between deadline, budget, and the required level of quality of service. The thesis demonstrates the capability of economic-based systems for wide-area parallel and distributed computing by developing users’ quality-of-service requirements-based scheduling strategies, algorithms, and systems. It demonstrates their effectiveness by performing scheduling experiments on the World-Wide Grid for solving parameter sweep—task and data parallel—applications. This paper focuses on introduction, grid definition and its evolution. It covers about grid characteristics, types of grids and an example describing a community grid model. It gives an overview of grid tools, various components, advantages followed by conclusion and bibliography.

4

1.

to provide a virtual plat-

INTRODUC-

form for computation

TION

and data management in

This The Grid

the same way that the

unites servers and stor-

Internet integrates re-

age into a single system

sources to form a virtual

that acts as a single

platform for informa-

computer - all your ap-

tion. The Grid is the

plications tap into all

computing

your computing power.

ture that will provide

fully utilized and spikes

the electronic. Grid in-

in demand are met with

frastructure will provide

ease. This Web site by

us with the ability to dy-

Oracle

namically link together

brings you the resources

resources

you need to evaluate

distributed applications.

when you are.

Grid is a type of par-

2. THE GRID

allel and distributed sys-

The Grid is the and

tem that enables the

data

sharing, selection, and

management infrastruc-

aggregation

ture that will provide

graphically

the electronic underpin-

of

geo-

distributed

"autonomous" resources

ning for a global society

dynamically at runtime

govern-

ment, research, science and entertainment, integrate

en-

resource-intensive, and

gies. The Grid is ready

business,

an

execution of large-scale,

option of grid technolo-

in

as

semble to support the

your organization's ad-

computing

data

management infrastruc-

Hardware resources are

sponsored

and

depending

on

their

availability,

capability,

performance, cost, and

networking,

users' quality-of-service

communication, compu-

requirements.

tation and information

5

3.

systems. However, in

BEGINNINGS

practice, the Grid has

OF THE GRID

been utilized more as a

Parallel com-

platform for the integra-

puting in the 1980s fo-

tion of loosely coupled

cused researchers’ ef-

applications

forts on the develop-

might be running in par-

grams and architectures

allel on a low-latency

that supported simultan-

parallel machine – and

eity. During the 1980s

for linking disparate re-

and 1990s, software for

sources (storage, com-

parallel computers foon

powerful

putation,

providing

tion and distribution –

for managing commu-

two fundamental con-

nication between pro-

cepts in Grid Comput-

cessors, and develop-

ing.

ment and execution en-

The first modern

vironments for parallel Successful

application

paradigms

were

developed

Grid is generally considered to be the information

to

ing infrastructure and

potential of shared and

applications for the I-

distributed memory ar-

WAY provided a semin-

chitectures. Initially it

ar and powerful experi-

was thought that the

ence for the first genera-

Grid would be most

tion of modern Grid re-

useful in extending par-

searchers and projects.

computing

This was important, as

paradigms from tightly

the development of Grid

coupled clusters to geographically

wide-area

year (IWAY). Develop-

leverage the immense

allel

visualization,

instruments). Coordina-

mechanisms

machines.

some

components of which

ment of algorithms, pro-

cused

–

research requires a very

distributed

6

different focus than dis-

t

tributed computing re-

r

search. Grid research

a

focuses on addressing

l

the problems of integra-

i

tion and management of

z

software.

I-WAY

a

opened the door for

t

considerable activity in

i

the development of Grid

o

software.

n

4. GRID COMPUT-

;

ING

CHARACa

TERSTICS An

n

enterprise-comput-

d

ing grid is characterized by three primary fea-

•

tures •

y

D

n

i

a

v

m

e

i

r

s

s

m

i

Diversity:

t

A typical com-

y

puting grid consists of

; •

D

many hundreds of man-

D

aged resources of vari-

e

ous

c

kinds

including

servers, storage, Data-

e

base Servers, Applica-

n 7

tion Servers, Enterprise

however, the systems

Applications, and sys-

and applications need to

tem services like Direct-

be able to flexibly adapt

ory Services, Security

to

and Identity Manage-

For instance, with the

ment Services, and oth-

late binding nature and

ers. Managing these re-

cross-platform

sources and their life

ties of web services, an

cycle is a complex chal-

application deployed on

lenge.

the grid may consist of a

Decentralization:

constantly changing set

changing

demand.

proper-

Traditional dis-

of components. At dif-

tributed systems have

ferent points in time,

typically been managed

these components can

from a central adminis-

be hosted on different

tration point. A comput-

nodes in the network.

ing grid further com-

Managing an applica-

pounds these challenges

tion in such a dynamic

since the resources can

environment can be a

be even more decentral-

challenging

ized and may be geo-

ing.

graphically

5. A COMMUNITY

distributed

across many different

OF GRID MODEL

data centers within an

Over

enterprise.

the

last

decade, the Grid com-

Dynamism:

munity has begun to

Components of a traditional

undertak-

converge on a layered

application

model that allows devel-

typically run in a static

opment of the complex

environment without the

system of services and

needing to address rap-

software required to in-

idly changing demands.

tegrate Grid resources.

In a computing grid,

The Community Grid

8

Model (a layered abstraction of the grid) being

developed

loosely

in

a

coordinated

manner throughout academia and the commercial sector. The bottom hori-

Figure 1: Layered archi-

zontal layer of the Com-

tecture of the Com-

munity Grid Model con-

munity Grid Model.

sists of the hardware re-

The next hori-

sources that underlie the

zontal layer (common

Grid. Such resources in-

infrastructure) consists

clude computers, net-

of the software services

works, data archives, in-

and systems, which vir-

struments, visualization devices

and

so

tualized the Grid. The

on.

key concept at the com-

Moreover, the resource

mon infrastructure layer

pool represented by this

is

layer is highly dynamic,

ment

both as a result of new

software,

Grid as a unified virtual

to the mix and old re-

platform and provide

sources being retired,

the target for more fo-

and as a result of vary-

cused software and ap-

ing observable perform-

plications.

ance of the resources in

The next hori-

the shared, multi-user of

on

agree-

which will represent the

resources being added

environment

community

zontal layer (user and

the

application-focused

Grid.

Grid middleware, tools and services) contains software packages built atop the common infra-

9

structure. This software

grid is most commonly

serves to enable applica-

used with large numbers

tions to more product-

of desktop machines.

ively use Grid resources

Machines are scavenged

by masking some of the

for

complexity involved in

cycles and other re-

system activities such as

sources. Owners of the

authentication,

desktop machines are

file

transfer.

6.

available

CPU

usually given control

TYPES

over

OF

when their re-

sources are available to

GRID

participate in the grid.

Grid computing can be

Data grid

used in a variety of

A data grid is re-

ways to address various

sponsible for housing

kinds of application re-

and providing access to

quirements. Often, grids

data across multiple or-

are categorized by the

ganizations. Users are

type of solutions that

not

they best address. The

concerned

with

where this data is loc-

three primary types of

ated as long as they

grids are

have access to the data. For example, you may

Computational grid

have two universities

A computational grid is

doing life science re-

focused on setting aside resources

search,

specifically

each

with

unique data. A data grid

for computing power. In

would allow them to

this type of grid, most

share their data, manage

of the machines are

the data, and manage se-

high-performance serv-

curity issues such as

ers. Scavenging

who has access to what

grid

data.

A scavenging

10

Another

com-

topology in or-

mon distributed com-

der to share and

puting model that is of-

collaborate.

ten associated with or

Many Grid dir-

confused

ectory

with

Grid

services

computing is peer-to-

implementations

peer computing. In fact,

are based on past

some consider this is

successful mod-

another form of Grid

els,

computing.

LDAP,

DNS,

7. THE KIND OF

network

man-

GRID TOOLS

agement

proto-

ing services.

components include file •

systems, schedulers and

is

Schedulers

like Grid FTP.

load

ser-

Systems

on a grid must capable

services

are

available

to

provide

this

function

and Sched-

ulers ensure that jobs are completed in some order

them. In short, systems

must be able to define

and

balancers

more.

of

discovering what

Grid

maximizing

efficiency.

file transfer mechanisms

be

balancers.

benefits of a grid

tificate authorities, and

Directory

and

One of the main

curity applications, cer-

vices.

Schedulers load

managers,

messaging systems, se-

•

as

cols, and index-

Infrastructure

resource

such

(priority,

deadline,

ur-

gency,

in-

for

stance) and load

(and

balancers

monitor) a grid’s

11

dis-

•

tribute tasks and

resources -- but

data

it can mean a lot

manage-

ment across sys-

more.

tems to decrease

stance, message

the

integrity

chance

of

inand

bottlenecks.

message confid-

Developer tools.

entiality are cru-

Every arena of

cial to financial

computing

and

en-

healthcare

environments

deavor requires tools that allow

8.GRID COMPON-

developers

ENTS:A

HIGH

LEVEL

PER-

succeed. for

to Tools

grid

velopers on

de-

SPECTIVE

focus

Depending

different

niches

expected use, some of these components may

nications, envir-

or may not be required,

onment control),

and in some cases they

and range from

may be combined to

utilities to full-

form a hybrid compon-

blown APIs.

ent.

Security. Secur-

8.1. Portal/user inter-

ity in a grid environment

face

can

Just as a con-

mean authentica-

sumer sees the power

tion and author-

grid as a receptacle in

ization -- in oth-

the wall, a grid user

er words, con-

should not see all of the

trolling who/what

on

the grid design and its

(file

transfer, commu-

•

For

complexities

can

computing

access a grid’s

of

the

grid.

Al-

though the user inter12

face can come in many

security, including au-

forms and be applica-

thentication, authoriza-

tion-specific.

grid

tion, data encryption,

portal provides the in-

and so on. The Grid Se-

terface for a user to

curity

launch applications that

(GSI) component of the

will use the resources

Globus Toolkit provides

and services provided

robust security mechan-

by the grid. From this

isms. The GSI includes

perspective,

an

A

the

user

Infrastructure

OpenSSL

imple-

sees the grid as a virtual

mentation.

computing resource just

provides a single sign-

as

of

on mechanism, so that

power sees the recept-

once a user is authentic-

acle as an. interface to a

ated, a proxy certificate

virtual generator.

is

the

consumer

created

when

It

and

performing

also

used ac-

tions within the grid. When designing your grid environment, you may use the GSI sign-in to grant access to the portal, or you may have your own security for the portal. The portal

Figure 2: Pos-

will then be responsible

sible user view of a grid

for signing in to the

8.2. Security

grid, either using the

A major require-

user's credentials or us-

ment for Grid comput-

ing a generic set of cre-

ing is security. At the

dentials for all author-

base of any grid envir-

ized users of the portal.

onment, there must be mechanisms to provide

13

This service is called the

Grid

Information

Service (GIS), or more commonly the Monitoring and Discovery Service (MDS). This service provides information about the available resources Figure 3: Security grid

in

within

the

grid and their status. A

a

broker service could be

environ-

developed that utilizes

ment

MDS.

8.3. Broker Once authenticated, the user will be launching an application. Based on the application, and possibly on

other

parameters

provided by the user, the

Figure 4: Broker service

next step is to identify

8.4. Scheduler

the available and appro-

Once the resources

priate resources to use

have been identified, the

within the grid. This

next logical step is to

task could be carried out

schedule the individual

by a broker function.

jobs to run on them. If

Although there is no

sets of stand-alone jobs

broker implementation

are to be executed with

provided

no

by

Globus,

interdependencies,

there is an LDAP-based

then

information

scheduler may not be

service.

14

a

specialized

required. However, if

might be used to sched-

you want to reserve a

ule work to be done on

specific resource or en-

a

sure that different jobs

cluster's

within the application

would handle the actual

run concurrently (for in-

scheduling of work on

stance, if they require

the cluster's individual

inter-process

nodes.

commu-

cluster,

while

the

scheduler

nication), then a job scheduler

should

be

used to coordinate the execution of the jobs. The

Globus

Toolkit

does not include such a scheduler, but there are several schedulers available that have been tested with and can be used in a Globus grid

Figure 5:

environment. It should

Scheduler

also be noted that there could be different levels of schedulers within a

8.5. Data management

grid environment. For instance, a cluster could be

represented

single

resource.

as

If any data -- in-

a

cluding

application

The

modules -- must be

cluster may have its

moved or made access-

own scheduler to help

ible to the nodes where

manage the nodes it

an

contains. A higher-level

will execute, then there

scheduler

needs to be a secure and

(sometimes

called a meta scheduler)

application's

reliable

15

method

jobs

for

moving files and data to various nodes within the grid.

The

Globus

Toolkit contains a data management component that provides such services. This component, know as Grid Access to Secondary

Storage Figure 6:

(GASS), includes facilData

ities such as GridFTP.

ment

GridFTP is built on top

8.6. Job and resource

of the standard FTP pro-

management

tocol, but adds addition-

With all the oth-

al functions and utilizes

er facilities we have just

the GSI for user authen-

discussed in place, we

tication and authoriza-

now get to the core set

tion. Therefore, once a

of services that help

user has an authentic-

perform actual work in

ated proxy certificate,

a grid environment. The

he can use the GridFTP

Grid Resource Alloca-

facility to move files

tion Manager (GRAM)

without having to go

provides the services to

through a login process

actually launch a job on

to every node involved.

a particular resource,

This facility provides

check its status, and re-

third-party file transfer

trieve its results when it

so that one node can initiate

a

between

file two

manage-

is complete.

transfer other

nodes.

16

9. ADVANTAGES Grid computing is about getting computers to work together. Almost every organization is sitting on top of enormous, unused computing capacity, widely distributed. Mainframes are idle 40% of the time With Grid computing, businesses can optimize Figure 7:

computing and data re-

Gram

sources, pool them for large

capacity

work-

8.7 Job flow in a grid

loads, share them across

environment

networks, and enable

Enabling an ap-

collaboration.

Many

plication for a grid en-

consider Grid comput-

vironment, it is import-

ing the next logical step

ant to keep in mind

in the evolution of the

these components and

Internet, and maturing

how they relate and in-

standards and a drop in

teract with one another.

the cost of bandwidth

Depending on your grid

are

implementation and ap-

mentum we're experien-

plication requirements,

cing today. Virtualiza-

there are many ways in

tion of the computing

which these pieces can

environment .

be put together to create

10.

a solution.

fueling

LLANGES GRID

17

the mo-

CHANOF

A word of cau-

11. CONCLUSION

tion should be given to

Grid computing in-

the overly enthusiastic.

troduces a new concept

The grid is not a silver

to IT infrastructures be-

bullet that can take any

cause it supports distrib-

application and run it a

uted computing over a

1000

faster

network of heterogen-

without the need for

eous resources and is

buying any more ma-

enabled by open stand-

chines or software. Not

ards. Grid computing

every application is suit-

works to optimize un-

able or enabled for run-

derutilized

ning on a grid. Some

decrease

kinds

applications

penditures, and reduce

simply cannot be paral-

the total cost of owner-

lelized. For others, it

ship. This solution ex-

can take a large amount

tends beyond data pro-

of work to modify them

cessing and into inform-

to

faster

ation management as

throughput. The config-

well. Information in this

uration of a grid can

context covers data in

greatly affect the per-

databases,

formance,

reliability,

storage devices. In this

and security of an or-

article, we outline po-

ganization's computing

tential problems and the

infrastructure. For all of

means of solving them

these reasons, it is im-

in a distributed environ-

portant for us to under-

ment. .

times

of

achieve

resources, capital

files,

ex-

and

stand how far the grid has evolved today and

12.

which features are com-

BIBLIO-

GRAPHY

ing tomorrow or in the

[1]

distant future

www.ibm.-

com/grid/index.html 18

[2] Foster, I. and Kesselman,

C.

(eds) (1999) The Grid: Blueprint for a New Computing

Infra-

structure..

San

Francisco,

CA:

Morgan Kaufmann [3] Berman, F., Fox, G. and Hey, T. (2003) Grid Computing: Making

the

Global

Infra-

structure a Reality.

Chichester:

John Wiley & Sons. [4] Web Site associated

with

book,

Grid

Computing: Making

the

Global

Infra-

structure a Reality, http://www.grid2002.org.

19