grid computing esis analysis of its impact on business and the shift to a services infrastructure

GRID COMPUTING:
ESI’s ANALYSIS OF ITS IMPACT ON
BUSINESS AND THE SHIFT TO A SERVICES
INFRASTRUCTURE

Robert Cohen, Ph.D., Fellow
Economic Strategy Institute


National Academy of Sciences GUIRR
February 24, 2004


WHAT IS DRAMATICALLY NEW
ABOUT ABOUT GRIDS?
a
a

a

a

GRIDS BEGIN AS A WAY TO ACCELERATE COMPUTING
THEY EVOLVE TO A TECHNOLOGY THAT MAKES
APPLICATIONS INFRASTRUCTURE AWARE
THIS IMBEDS WEB SERVICES WITHIN A NEW INTELLIGENT
LAYER BETWEEN APPLICATIONS AND COMPUTE/STORAGE
RESOURCES
THIS CREATES THE “SERVICES GRID”
• PRIORITIZING & AUTOMATING THE USE OF
COMPUTE/STORAGE RESOURCES AND

• AUTOMATING COMPLEX BUSINESS PROCESSES

2


WHY GRIDS ARE REVOLUTIONARY
a

THEY PROVIDE A STRUCTURE FOR THE NEW
SERVICES ECONOMY - INCLUDING SERVICES AND
CONTROL IN MANUFACTURING
• LET CORPORATIONS MANAGE CRITICAL STANDARDS ACROSS
THE FIRM: CREDIT RISK, SIX SIGMA
• SUPPORT COLLABORATIVE PRODUCT DEVELOPMENT IN THE FIRM
& WITH PARTNERS
• ANALYTICAL CAPABILITIES OF A NEW ORDER REDUCE FINANCIAL
RISK, PROVIDE A TYPE OF “SIX SIGMA” FOR SERVICE FIRMS
• CREATE WAY TO SUPPORT SERVICE-BASED PRODUCTS
3


GRIDS: BREAKING INTO THE
SERVICES SIDE OF BUSINESS
a

IDC:
– Grids hold significant potential, especially if the technology can
break out from the HPC marketplace and capture the attention of
commercial users.

– If this "chasm" can be crossed, revenue sources will largely shift
from hardware in the HPC market to a more software and services
based model in the commercial market

a

THE MANAGEMENT AND SERVICES SIDE OF GRID IS FAR
GREATER THAN DIRECT COMPUTER SERVICES
– HUGE REDUCTION IN RISK FACING BANKS, CORPORATIONS
– GREATER CONTROL OF PRODUCT DEVELOPMENT, SUPPLY CHAINS,
FINANCIAL PERFORMANCE
4


What is a Grid? - Ian Foster, “What is the Grid? A Three
Point Checklist”
a

1. Coordinates resources that are not subject to centralized control
…
– integrates and coordinates resources and users that live within different control domains—for
example, the user’s desktop vs. central computing;
– It addresses the issues of security, policy, payment, membership, and so forth that arise in these
settings. Otherwise, we are dealing with a local management system.

a

2. Uses standard, open, general-purpose protocols and interfaces
– built from multi-purpose protocols and interfaces that address such fundamental issues as
authentication, authorization, resource discovery, and resource access.

– these protocols and interfaces must be standard and open. Otherwise, we are dealing with an
application specific system.

a

3. Delivers nontrivial qualities of service.
– its constituent resources are to be used in a coordinated fashion to deliver various qualities of service,
relating to response time, throughput, availability, and security, and/or co-allocation of multiple
resource types to meet complex user demands
– the utility of the combined system is significantly greater than that of the sum of its parts.

5


30000

Does not include downloads from:
NMI, UK eScience, EU Datagrid,
IBM, Platform, etc.

GT 2.0
Released

25000

Physiology of the Grid
Paper Released
20000

Anatomy of the Grid

Paper Released
NSF & European Commission
Initiate Many New Grid Projects
DARPA, NSF,
and DOE
begin
funding Grid
work NASA begins
funding Grid work,
DOE adds support

Significant
Commercial
Interest in
Grids

15000

10000

GT 1.0.0
Released

Early Application
Successes Reported
5000

D o w n lo a d s p e r M o n th fro m ftp .g lo b u s .o rg

Globus Toolkit® History

0
1997

1998

1999

2000

2001

2002

6


Globus Open Source Grid Software
G
T
4

G
T
3
G
T
2
G
T

3
G
T
4

Community
Scheduler
Framework
[contribution]

Delegation
Service

Python WS Core
[contribution]
C WS Core

Community
Authorization
Service

OGSA-DAI
[Tech Preview]

WS
Authentication
Authorization

Reliable
File

Transfer

Grid
Resource
Allocation Mgmt
(WS GRAM)

Monitoring
& Discovery
System
(MDS4)

Java WS Core

GridFTP

Grid
Resource
Allocation Mgmt
(Pre-WS GRAM)

Monitoring
& Discovery
System
(MDS2)

C Common
Libraries

Pre-WS

Authentication
Authorization

Web
Services
Components

Components

Replica
Location
Service

XIO

Credential
Management

Security

Data
Management

Non-WS

Execution
Management

Information
Services

Common
Runtime


Services Oriented Infrastructure as the
Next Phase in the Evolution of Grids

Infrastructure is Globus
that is “completely”
Web Service-ized

Source: Rich Miller, Univa

The SOI links to applications using
Grid Services, Applications are
linked by Web Services.

8


What is a Cluster vs. a Grid?

9
Gentzsch, “Grid Computing: A New Technology for the Advanced Web”


GRID: FLEXIBLE, HIGH-PERFORMANCE
ACCESS TO ALL SIGNIFICANT RESOURCES
On-demand creation of powerful virtual computing systems

Source: Dietmar Erwin, “UNICORE and EUROGRID: Grid Computing in Europe,” EUROGRID 10
and UNICORE


GRIDS LINK GLOBAL OPERATIONS:
COORDINATING SERVICES

11


GRIDS IN THE CORPORATION:
COMPUTE RESOURCES, STORAGE,
CONTROL

Wolfgang Gentzsch, Director Grid Computing, Sun Microsystems, “Sun Grid Computing: Vision, Strategy, Products,”

12


13


Globus as
Service-Oriented Infrastructure
User
Application
Tool
Uniform interfaces,
security mechanisms,

Web service transport,
monitoring
GRAM

User
Application
Tool

User
Application
Reliable
File
Transfer
MDSIndex

User Svc

Host Env

User Svc

Host Env
MyProxy

GridFTP

DAIS
IBM

IBM

IB M

IB M

Computers

Specialized
resource

Storage

Database
14


Service Oriented Computing – An Ecosystem that Manages
Change as a Competitive Weapon
Utility
Grids

Metering

Provisioning

Provisioning

Performance Management

Performance Management

Processing
Networks

SLA

Virtual
Resource
Market
Supply

Storage

Data Grids

Accountability

Demand

Compute Grids

Functionality

Charge-Backs

Governance
Users
Agility, Quality,
Cost, Risk

Services
Web Services
Semantics

Governance

Applications
Business Logic,
Workload Management

Policies
Auctions, Commodity
Markets, Rule Based, etc

SOA
Service Oriented Architecture
© Copyright 2004 CSFB

Virtual Resources Physical Resources
Process/Data Migration,
Clustering, SSI, CFS

SOI

Best of Breed,
Commodity H/W

Service Oriented Infrastructure

15

ECONOMIC RATIONALE OF ANALYSIS
a

GRIDS ARE TECHNOLOGICAL INNOVATIONS THAT LOWER THE
OPERATING COSTS OF ADOPTER FIRMS.
• BY REDUCING COSTS, ADOPERS BECOME MORE EFFICIENT. THEY BRING
PRODUCTS TO MARKET AT LOWER COST. THIS FREES UP CAPITAL FOR NEW
INVESTMENT BY THE EARLY ADOPTER INDUSTRIES.
• WHEN EARLY ADOPTER FIRMS PRODUCE CHEAPER GOODS THAN THEIR
COMPETITORS, THEY ARE ABLE TO SELL MORE AND INCREASE PROFITS.
– THEY GROW FASTER AND “CROWD OUT” LESS DYNAMIC SECTORS.
– THEY ATTRACT MORE INVESTMENT THAN SLOWER GROWING SECTORS.

a

EARLY ADOPTERS GAIN KEY ADVANTAGES THAT ARE REINFORCED
BY ADDING MORE OPERATING EFFICIENCIES.
– THEY REDESIGN THE WAY THEY DEVELOP PRODUCTS AND GET THEM TO MARKET
FASTER.
16
– AS A RESULT, THEY BECOME EVEN MORE DYNAMIC. A “VIRTUOUS CYCLE.”


MAIN INDUSTRY FINDINGS
a

FROM 2003 AND 2010, 8 EARLY ADOPTER INDUSTRIES WILL INCREASE THEIR
SALES DRAMATICALLY DUE TO GRIDS AND WEB SERVICES.

• THE LARGEST GAINS WILL BE IN COMPUTERS (36%, WITH 19% OF THE GAINS DUE
TO GRIDS), AUTOS (19%, WITH 12% OF THE GAINS DUE TO GRIDS) AND
PHARMACEUTICALS (13%, WITH 7% OF THE GAINS DUE TO GRIDS).

a

OVER THIS PERIOD, EARLY ADOPTER INDUSTRIES WILL EXPERIENCE MAJOR
GAINS IN PRODUCTIVITY DUE TO GRIDS AND WEB SERVICES.
• THE GREATEST PRODUCTIVITY GAINS WILL BE IN FINANCIAL SERVICES (26%)
AUTOS AND PHARMACEUTICALS (25%) AND HOSPITALS (24%) AND AEROSPACE
(20%).

a

MANY GRIDS WILL BE MULTI-LOCATION ENTERPRISE GRIDS BY 2010. THIS
WILL RESULT IN GREATER SPENDING ON BROADBAND ACCESS.
• THE AUTO INDUSTRY WILL INCREASE THIS SPENDING FROM AN EXPECTED $0.5
BILLION TO ABOUT $2.3 BILLION ( 1996 DOLLARS).

17


Percentage Change in US Industries' Sales due to the Adoption
of Grids and Web Services, 2002-2010
40.0%
35.0%
Percentage
30.0%

Change over

25.0%
Forecast for 2010
20.0%
15.0%
10.0%
5.0%
0.0%
-5.0%

Changes
due to the
Adoption
of Web
Services
Changes
due to the
Adoption
of Grids
and Web
Services
18


EXAMPLES OF ADOPTION FIGURES FOR FINANCIAL
SERVICES AND AUTOS

The main difference is the more rapid deployment of Enterprise grids in financial services
and the earlier development of partner grids. Some of the biggest banks and brokerage firms
were making dramatic investments in clusters, grids and Web services in 2003. They were
19

the first to adopt blade servers in large numbers.


GRID ADOPTION - PHARMAS,
AEROSPACE

20


GRID ADOPTION - HEALTH CARE

21


Communications
Services

Hospitals

Financial
Services

Aerospace

Autos

Pharmaceuticals

Elec.
Components

Computers

Percentage Change over Forecast
Level in 2010

Changes in U.S. Industry Productivity Due to the
Adoption of Grids and Web Services, 2002-2010

30%
25%
20%
15%
10%
5%
0%

22


GRIDS RAISE PRODUCTIVITY IN AUTOS AND
FINANCE
Labor productivity - Autos
245

Comparing Adoption of Grids (Blue)
and Baseline Forecast to 2010

Labor productivity - Financial Services
87.9

187

65.9

130

Comparing Adoption of Grids (Blue)
and Baseline Forecast to 2010

43.9

1990
baseline

1995

2000

Grid Adoption

2005

2010 1990
baseline

1995

2000

2005

2010

Grid Adoption

23


AEROSPACE AND HOSPITALS HAVE
LOWER PRICES
Prices - Aerospace
Comparison of baseline and Adoption of Grids

Prices - Hospitals

1.39

Comparison of baseline and Adoption of
Grids
2.27

1.23

1.75

1.06

1.22

1990
baseline

1995

2000
Adoption of Grids

2005

2010

1990
base

1995

2000

Adoption of Grids

2005

2010
24


ESI: SPENDING ON GRIDS TO BE $7.5 BN
SOFTWARE, $35 BN ALL COSTS IN ‘07
Cost of Deploying Grids, Software Costs Only, 20032010

Spending to Deploy Grids, All Costs, in Millions of
Dollars, 2003-2010

160000
AUTOS
HEALTH CARE
AEROSPACE
PHARMACEUTICALS
FINANCIAL SVCS
COMPUTERS
SEMICONDUCTORS

25000
20000
15000
10000
5000
0
2003-2005

Millions of Dollars

180000

30000
Millions of Dollars

35000

140000

PHARMACEUTICALS
AUTOS
HEALTH CARE
AEROSPACE
FINANCIAL SVCS
COMPUTERS
SEMICONDUCTORS

120000
100000
80000
60000
40000
20000

2006-2007

2008-2010

0
2003-2005

2006-2007

2008-2010

THESE ESTIMATES ARE ABOUT 60% MORE THAN THE INSIGHT ESTIMATES FOR 2008 (SOFTWARE
ONLY) AND ABOUT 5 TO 6 TIMES THE INSIGHT ESTIMATES IF ALL COSTS INCLUDING

COMMUNICATIONS COSTS ARE INCLUDED.
THE SPENDING BY THE FINANCIAL SERVICES SECTOR DRIVES MUCH OF THE SPENDING TO DEPLOY
GRIDS THROUGH 2010. WE ARE RECHECKING THIS ASSUMPTION TO SEE IF IT NEEDS TO BE MODIFIED.

25