Transcript Emerging Technologies

NBA 600: Session 26
New Technologies
24 April 2003
Daniel Huttenlocher
About The Presentations
 Each group should have one presenter
– Everyone prepared to answer questions
– Spend about 8-10 mins presenting (~ 5 slides)
– Make sure to set context for your industry and
technology
– Send presentation by 2:30 Tues; use my laptop
 Each of you will be given one-page sheets
to comment on each presentation
– Must be handed in to me by next Friday, May 2
• Will count in your class participation grade
– Will help inform my assessments of presenters
2
Today’s Class
 Some new or emerging technologies
– In context of what an IT-savvy manager
should know or know how to learn (last time)
 Finish discussion of Web services
– Software platforms for delivering them
 IT as a service
– From ASP’s to outsourcing to “utilities”
 Grid computing
– Large scale networked resource sharing
 Autonomic computing
– “Self healing” systems
3
Web Services Software
 Two application development frameworks
make easier to deploy Web services
– J2EE from Sun, based on Java
• Also supported by IBM, Oracle and BEA
• Proprietary extensions from each vendor
– .net from Microsoft, based on CLR
• CLR: common language runtime
 Language independent but primarily new
language C# and Visual Basic
• Wide adoption in Microsoft developer community
 In practice, many are using both
4
Recent Study
 Gartner survey from September ’02
– 44 consulting and systems integration firms
– Reported in Information Week, 2/5/03
 Top 3 platforms targeting for Web services
–
–
–
–
58% .net
40% IBM WebSphere (J2EE)
31% Oracle (J2EE)
Sun fourth place
 Survey of 140 companies similar results
– Smaller companies more likely to use .net
– Larger more likely to use J2EE or both
5
What’s Meant by Web Services
 Most companies still using Web services
within the enterprise
– Some starting to offer services to outsiders
• Beyond technology leaders like FedEx, Google,
Amazon, Ebay
 Generally using XML for inter-system
communication over HTTP
 Usage of SOAP and WSDL still low
– In Feb. 2002 was “miniscule”
– Currently around 20% report using at least
one
6
Full Web Services Architecture
Discovery
Agency
Find
(Using WSDL)
Publish
(Using WSDL)
IP Network
Service
Requestor
Interaction
(Using SOAP)
Service
Provider
7
Today’s Web Services Architecture
IP Network
Service
Requestor
Interaction
(Using XML over HTTP)
Service
Provider
8
Web Services Stack
 How many layers deployed to meet
today’s needs versus tomorrow’s
Tool
Layer
Needed
UDDI
Service
Discovery
If services to be found
dynamically
WSDL
Service
Description
If services to be found
dynamically
SOAP
XML-based
messaging
If transport to change
or services found dyn.
HTTP, FTP,
SMTP, etc.
Transport
Always (and XML
messages)
9
Management Decisions
 Vendors and platforms
– Unix/Java or Microsoft shop (often both)
• J2EE or .net (or both)
– Currently using
• Other trends driving these choices within firm
 What are potential customers using and
how much influence over their choices
– Or compatibility across vendors
 How far up the Web services stack
– Is minimum for the business purpose
– Is desirable for future needs
10
IT Products vs. Services
 Application software long been offered
either for sale or for rent/use
– Purchase model vs. application service
provider (ASP) model
– Consumer example: buy Turbotax vs. use on
Web
• No need to install, or even have own computer
– Corporate examples: HR, payroll, email and
other core applications
• Again, no installation, maintenance, etc.
• Easily quantifiable costs if long term contract
11
ASP Model
 Has tended to be more successful with
smaller businesses
– Less ability to have/afford adequate in-house
technical expertise
 Has been successful in data delivery
– Particularly financial data: Reuters, Bloomberg
 Larger firms still prefer purchase/license
– Critical systems under own control
• Vs. ASP as well as network to get there
– Data privacy/security/integrity issues
– Can amortize IT staff over larger base
12
Computing as Utility
 Being pushed by IBM, HP, and to lesser
degree Sun
– Companies should not make large capital
investments in IT infrastructure
• Computers, networks, operating systems and
“middleware” software
• Inefficient use of resources
– Instead view as utility like electricity
• Pay for what you use
– IBM: on-demand computing
– HP: adaptive infrastructure
– Sun: N1, computing to n-th degree
13
Utility Computing History
 IBM, EDS and others provided “computer
service bureaus”
– Access to mainframe computing resources
• Pre-internet, mainly over leased-line networks
 During dot-com boom many managed
hosting services provided Web servers
– Still true, but more acting as data center space
– Varieties of service levels
• Complete Web server
• Data center space and networking
• Just data center space
14
On-Demand Computing
 IBM positioning as a thought leader
– Palmisano giving many speeches on the topic
 View as tightly coupled with
– Commoditization of computing and networking
• Linux on X
– Web services model
• General application platforms (WebSphere a
leading one)
• Standard interfaces (XML, SOAP, etc.)
– Grid computing
• Managing large networks of resources
15
IDC Survey
 Survey of 34 companies
– Reported in CNET News Feb 14, ’03
 Found IBM and HP reported top choices
for providing utility computing services
 19 of 34 wouldn’t want such services
offsite or shared with other companies
 Almost half interested in “private utility”
– In-house use of software and services to more
effectively use computing resources
 Seeking a lot
– Short contracts with savings of 28% or more
16
Utility Computing Concerns
 Similar to all support or service
outsourcing projects
–
–
–
–
Financial stability of provider
Safety of data
Accessibility/reliability/service level
Escalating costs
 On other hand desperately looking to
lower basic IT costs
– Networking, storage, computing, operating
systems and infrastructure software
– Have been substantial outsourcing deals
17
IT Infrastructure Outsourcing
 Financial services industry undertaking
large outsourcing of basic IT services
– JP Morgan Chase 7yr $5B deal with IBM
• IBM absorb 4,000 JPMC employees/contractors
• Plan to eliminate half of 16,000 servers globally
• Consolidate 37 independent networks into one
• Annual cost reductions over contract life
–
–
–
–
Deutsche Bank 10yr $2.6B with IBM
B of A 10yr $4.5B with EDS
ABM Amro 5yr $1.3B with EDS
CIBC 7yr $1.5B with HP
18
Grid Computing
 Term coined in mid ’90’s
– For distributed computing infrastructure for
advanced science and engineering
 Has come to refer to many things
– globus.org grid toolkit used by many research
and academic organizations
• Open source (using own globus license)
• Also basis of solutions by vendors such as IBM
– Globus defines grid as “coordinated resource
sharing and problem solving in dynamic, multiinstitutional virtual organizations”
19
Grid Computing Toolkit
 Sharing computing and storage resources
– Across wide area networks
– Large number of machines
• Potentially different hardware/OS
– Machines possibly owned and administered by
different organizations
• With different goals and policies regarding
priorities of tasks on their machines
 Originally targeted at academic and
research projects – “supercomputing”
– Less attention to security/privacy
• Now focus at IBM and elsewhere
20
Autonomic Computing
 Computing systems are getting more
complex and inter-connected
– Maintenance and system administration issues
are reaching a crisis
 Idea of autonomic computing is to give
systems some degree of “self awareness”
– Ability to detect problems
– Ability to automatically correct problems
• Notify humans as appropriate
– Currently considerable human monitoring and
re-configuration to keep running
21
Status of Autonomic Computing
 Some large IT systems vendors are
moving in this direction
– Databases from IBM and Oracle doing more
self-tuning and self-analysis
• Traditionally have substantial human oversight
– IBM has broad autonomic computing focus
• Working to accelerate both business adoption
and new academic research
 Solutions range from technically feasible
today to very ambitious proposals
– Large scale research projects at several
universities, government funding
22