CERN School of Computing 2000 European Commission: DG Information Society 1 Distributed Applications [email protected] The opinions on this.

CERN School of Computing 2000CERN School of Computing 2000

European Commission: DG Information Society

1

Distributed Applications

[email protected]

The opinions on this presentation are those of the author and doesn’t reflect necessarily the opinion of the European Commission.

The primary source of data is quoted - which has in some cases been copied from secondary sources and has not always been checked .



2

CONTENTS* Research Trends - Active Networking

- Object /component based software- GRID Technologies

* Market Trends- Telecom Services- ASP- Community Computing

* A possible future scenario



3

ACTIVE NETWORKING

* Today’s network components are reprogrammable only by the vendor.

* The nodes in an Active Network are able to receiving and executing active packets for customising the network services.

By opening the network nodes Service Providers will be freed from :

* Vendor dependent implementations

* The time consuming standardization process

* Finding work arounds to implement functionality



4

Components of an AN

1. A programmable Networking Infrastructure* Open Programming Interfaces* Protocol stak versus protocol components* VANs

2. Active packets* Mobile Agents for service creation to customise the packet handling service related to a user connection

3. The programming language* JAVA



5

Mobile Agent Platforms differ on

* Operating Systems

* Architectural choices

Two main approaches

* MASIF (Mobile Agents System Interoperability Facility)

* FIPA (Foundations for Intelligent Physical Agents)



6

1. IST Programme Actions

Activities running : ANDROID, FAIN, AGENTLINK

Draft WP2001 : IV.2.2 Network and Services interoperability, Interworking and Management

OBJECTIVE : To increase network agility and functionality and to support service interworking and management

FOCUS : Active and dynamically reconfigurable network technologies, methods and tools



7

2. Object Oriented and Component Based distributed Software

* Development platforms : CORBA, JAVA

* Compound document Frameworks : OpenDOC, EJB

* Unified Modeling Language (UML)

The facto standards

Stable products in the market

Open Source products



8

Main Issues

* No product covers the complete development cycle: Analysis, Specs, Design, Programming, Testing, Maintenance

* Tools for automatic code generation from an UML model

* Defining what is a component

* Tools to build and administer the components library to ensure re-usability and maintenance



9

Main Issues (continued)

*Interoperability between different implementations

- Compliance Testing

- Performance Testing

* Inadequacy for time-critical applications



10

1. IST Programme Actions

Activities running : CORVAL2, DSE PLS Consulting Study

Draft WP2001 : IV.3.1 Software Architecture Focuses on models and notations for component

based software.

IV.2.1 Real-Time distributed Systems Focuses on models, technologies and tools



11

3. GRID TechnologiesA collection of middleware services to cluster through a high-speed network a wide variety of resources (computer, storage systems, data resources, and special devices) distributed geographically to enable the use of applications from anywhere without knowing where such applications are executing or where the data is located.

* Wide-Area distributed super-computing, GLOBUS, UNICORE, PROMENVIR/DISTAL

* High data throughputMONARC



12

3. GRID LAYERS* Computational GRID

* processing power* communication links* data storage

* Information GRID* access to information* analysis* visualization

* Knowledge GRID* data mining* machine learning* decision making support



13

Main Risks/Challenges

* to focus on the architecture no efforts on the development environments

* Standardisation. The GRID versus many GRIDS

* To provide for a collaboration environment



14

IST Programme Actions

Running activities : EUROGRID, DATAGRID June Workshop and Perrott’s Report

Draft WP2001 : CPA9: GRID Testbeds

* To validate not only technology but also to quantify and quality the benefits and risks from the user point of view.

* To develop GRID specific-Tool-kits

* Promotion of standards



15

Market Trends 1. Telecom Services

* Companies outsource operations : Call Centres127.000 agent positions will be outsourced by 2003 (EC)

3.5 million agents will be working in carrier-based call centres by 2005 (OVUM)

* Internet services are largely outsourced e-mail, Web Site design, access, e-commerce, etc.

* VoIP is taking-off and will make possible the convergence of the 2 types of services.



16

* Unified messagingThe convergence of voice, fax, e-mail and Web

* 3 Generation mobile (UMTS)800 million terminals will be used for data access by 2004(ARC group)

* Business are adopting new methods of location - independent working2 million teleworkers in 19974.6 million in 19989 millions in 1999 (EC)



17

2. Application Service Providers

An ASP manages and delivers application capabilities on a subscription basis to multiple users across a wide are network.

* remote access to hosted applications

* flat rate per user business model

* on his own predefined computing platform



18

Market predictions for ASP

* The European ASP will grow from $ 100 m in 2000 to $ 1.5bn in 2004 (Durlacher Research)

* The global ASP market will reach $ 5.1 bn by 2001 (Datamonitor)

* ASP services will soar from 889 m in 1999 to over $ 22.7 bn in 2003 (Gratner’s Dataquest)

* ASP market will reach $ 44 bn in 2004 of which $ 8.7 bn will be on the EU market (OVUM)



19

ASP Actors* global companies :

December 1999 : IBM’s ASP Prime program January 2000 : CHAI (Cisco Hosted Applications Initiative)June 2000 : CW + Microsoft + Compaq

Alliance for ASP services* start-ups

April 2000 : InforSense (Imperial College London)May 2000 : Red2plc (Catalyst Solutions)

* ASP Industry Consortium May 1999 : formed by 25 companies

Today : about 300 members



20

3. Community computingLinking millions of otherwise idle PCs worldwide to build a super-computer

* Mid 1999 SETI@home project - University of California in Berkeley 2 million people have downloaded the screen saver

* January 2000 Popular Power founded by a former MIT Lab researcher. An exchange point for buyers and setters of computing time.

* January 2000 Process Tree founded by a former Intergraph Staff member . More than 120.000 computers have signed-up



21

Meaningful research cannot be done in the ivory tower but has to relate in some way to reality and to the needs of society. In this sense, there is no pure science. Even scientists in abstract fields such as number theory or elementary particle theory should constantly think about the practical relevance of their research.Obviously, research funding agencies should not ask for immediate social benefits from research projects. They, as well as the scientist,have to look far ahead into the future.

Richard ErnstNobel Prize in Chemistry 1991



22

A possible future scenarioThere are Applications Portals allowing a user to select a set of application services irrespective of the platform or organization through which the service is provided.

These Portals provide the needed integration technology to front all the applications and to integrate with in-house systems. Confidentiality and integrity are ensured. They manage not only hardware resources but also license issues. They take responsibility for the resources consumed providing settlements between providers and a single invoice to the user. Credit-card checking and secure payments are incorporated.

As today’s ISP don’t have to own their network resources Application Portals don’t own the computer resources.

.



23

From an EC Report based on the European Information TechnologyObservatory (EITO) Report from 1999,

the EU software market is estimated in 186 Bn Euro * Software products : 38 Bn* Traded services (outsourced) : 68 Bn* Non-traded services (in-house) : 80 Bn

Application Portals will influence drastically

ARE WE GOING TO THE COMPUTERLESS ENTERPRISE ??ARE WE GOING TO THE COMPUTERLESS ENTERPRISE ??

CERN School of Computing 2000 European Commission: DG Information Society 1 Distributed Applications [email protected] The opinions on this.

Documents

european commission

information society

cern school of computing

information grid

tools slide

community computing

maintenance slide

java slide