43
19/5/08 TNC Evolution of research networks: quantity? quality? functionality? Paul VAN BINST Université Libre de Bruxelles

19/5/08 TNC Evolution of research networks: quantity? quality? functionality? Paul VAN BINST Université Libre de Bruxelles

Embed Size (px)

Citation preview

19/5/08 TNC

Evolution of research networks:quantity? quality? functionality?

Paul VAN BINST

Université Libre de Bruxelles

19/5/08 TNC

Personal/Historical Introduction

19/5/08 TNC

Why was I invited to talk to you today?

19/5/08 TNC

Why was I invited to talk to you today?

• Because:

– I am old enough?– I had a great career in networking?– I am a good old friend of TERENA?– I am a good speaker?

19/5/08 TNC

I am old enough?

• I started my career in 1967…

19/5/08 TNC

I am old enough?

• I started my career in 1967…

19/5/08 TNC

I am old enough?

• I started my career in 1967…

… and I stop next September

19/5/08 TNC

I had a great career in networking?

• I was probably part of the pioneers…

19/5/08 TNC

I had a great career in networking?

• I was probably part of the pioneers…

P. VAN BINST, R. VANDENBROUCKE

Development of a communication facility between two PDP‑11's and a DEC system 10.Société Belge de Physique, Réunion Scientifique Générale, 1980

M. GOOSSENS, J. TIBERGHIEN, P. VAN BINST, R. VANDENBROUCKE

Development of a file transfer capability between a PDP‑11 under RSX‑11M and a HP 1000 under RTE.DECUS Europe Symposium, 1982

19/5/08 TNC

I am a good old friend of TERENA?

• Definitely!– I started working in the early 80’s at the

European level (with ECFA) to create, finance and deploy international networking infrastructures for research

– I was at the first meeting where the concept of such as thing as TERENA was discussed

– I represented Belgium in RARE/TERENA for 10 years (1985 – 1995)

– I was in the executive committee for 6 years

19/5/08 TNC

Let me show you three souvenirs about the previous slide…

19/5/08 TNC

ECFA - European Committee for Future Accelerators

ECFA/82/60

Networks for High-Energy PhysicsAn Interim Report on Wide-Area Communications

ECFA Working Group on Data Processing Standards

Subgroup 5 (Links and Networks)August 1982

19/5/08 TNC

19/5/08 TNC

19/5/08 TNC

I am a good speaker ?

… I’ll leave it to you to decide … but I surely spoke a lot in my life!

19/5/08 TNC

I am a good speaker ?

… I’ll leave it to you to decide … but I surely spoke a lot in my life!

19/5/08 TNC

I am a good speaker ?

… I’ll leave it to you to decide … but I surely spoke a lot in my life!

19/5/08 TNC

I am a good speaker ?

… I’ll leave it to you to decide … but I surely spoke a lot in my life!

19/5/08 TNC

I am a good speaker ?

… I’ll leave it to you to decide … but I surely spoke a lot in my life!

19/5/08 TNC

Evolution of research networks(past, present and future)

• Quantity ?

• Quality ?

• Functionality ?

… an executive summary

19/5/08 TNC

Quantity (1)

• A long story of increasing data rates at decreasing prices …

• … which I am not going to tell today, or even summarize

19/5/08 TNC

Quantity (2)

• Two references:– An achievement : international 256 Kbps from

Belgium, 1987A. COHEN, C. HANON, T. MASSART, N. MEULEMANS, P. VAN BINST,

R. VANDENBROUCKEThe Helios‑B Project: Towards High Speed File Transfers by Satellite

Between Heterogeneous Computers.Proceedings of TENCON 87, IEEE

– A long fight : European 2 Mbps backboneP. VAN BINST (ed.)Proceedings of the Third Symposium on High Speed Networking for

Research in Europe, Brussels, February 1994.Computer Networks for Research in Europe, Elsevier, 25, 1994

19/5/08 TNC

Quantity (3)

• Executive summary :

– For a long time, the problem was price …– … then came the great 1988 – 1998

decade: de-monopolization, liberalization of the European telecom market …

– … which resulted in the price/performance of the networks we use today

19/5/08 TNC

Quality (1)

• The (old) concept of Quality of Service (QoS) has evolved considerably towards today’s Service Level Agreements (SLAs)

• QoS used to be defined and measured at the physical and network layers, with the main focus being put on Bit Error Rate (BER), as well as transit delay, jitter, etc. – in a world of virtual circuits / connection-oriented and broadcast networks– That is still the case today for most telephony,

radio and TV networks – with billions of users

19/5/08 TNC

Quality (2)• Internet and LANs (connectionless / datagram

networks) changed the situation because of the essentially unreliable (« best effort ») lower layers

• The practical questions of QoS moved to layers 4 (TCP or UDP) and higher (applications and services) – also due to the much improved performance of the lower layers

• AND: the networks are mostly used by people (human beings) for voice, music, radio and video communications, and this in fact allows to drop the « hard » requirement for QoS (the « what » is more important than the « how ») – we listen to the news, we watch the football game, …

19/5/08 TNC

Quality (3)

• Since a number years, the real basic answer to QoS questions at the lower layers is called « over-dimensioning » - exactly what we witness today in GEANT (or Belnet), as in most operators networks

• Precise QoS measurements in the lower layers are sometimes still carried out for industrial and professional users / contracts, but not for the vast majority of end users

19/5/08 TNC

Quality (4)

• Most users (also on NREN’s) don’t see the quality of the lower layers, which is very often masked by the unpredictability of Internet applications and services

• Just one example: in a recent series of long file downloads into my PC at ULB, at various times by day and night, from a major university, well connected to its NREN in a European capital city, the observed sustained data rate varied by a factor 40 ! (from 40 KBps to 1.6 MBps)– That’s like a car doing sometimes 160 km/h, sometimes

4 km/h !!!

19/5/08 TNC

Executive Summary (up to here)

• We have (more than) enough quantity

• So we don’t care too much about quality

• Especially if we love the functionality

19/5/08 TNC

Functionality (1)

• The real paradigm shift is that we are preferring functionality to quality

• The obvious best example of this behaviour is the SMS (more than 1 billion per day in the world) – a very basic service with a lousy ergonomy

• Our regular web attitude is that we don’t mind to wait, either a fraction of a second or a large number of seconds, as long as we like what we get

19/5/08 TNC

Functionality (2)

• All the same, we look at a poor definition TV broadcast or video in a small window on our computer screen while we have a high-quality, large TV set next door

• … and I will not comment on the phone or video conferencing applications tragedy…

19/5/08 TNC

Functionality (2)

• All the same, we look at a poor definition TV broadcast or video in a small window on our computer screen while we have a high-quality, large TV set next door

• … and I will not comment on the phone or video conferencing applications tragedy…

• Meanwhile, the world is changing and people using (non research) networks are getting better and better services, with good (sometimes guaranteed) quality and functionality – think eg. of the banking or health worlds…

19/5/08 TNC

A (Belgian) example of advanced functionality in the health world

The VIGITENS project, carried out by– St Peter hospital in Brussels– Hodie Vivere foundation– Belgacom– Tandberg– ULB/STC

Remote mentoring of emergency cases

19/5/08 TNC

Functionality (3)

• Do research networks need to develop more advanced functionality?

• We have already some good and performant services in the NREN world, such as CERTs or EDUROAM

• One of the nicest « new » things is the development of the grid paradigm, because it means networks, technology, services and management – let me steal a nice slide from the EC

19/5/08 TNC

e-Infrastructure - implementation blocks

networking

specific services

joint research activ

ities

GÉANT .

INFRASTRUCTURE

networking

specific services

joint research activ

ities

GRID .

INFRASTRUCTURE

19/5/08 TNC

BEgrid

I “must” have one slide about BEgrid!• Infrastructure

– > 800 CPUs– about 30 Tbyte of storage– > 600 valid certificates

• Usage in 2007– > 600000 jobs– > 150 years of CPU time

• Integrated in EGEE (gLite middleware), > 40000 CPUs• Integrated with NLgrid• Ready to be an NGI in the EGI structure

19/5/08 TNC

Functionality (4)

• But we are still far away from the « sexiness » of Google Earth, YouTube, ViaMichelin or DVB-H…

• How separate can (should?) NRENs be (or remain?) from other (public?) services?

19/5/08 TNC

A last example from Brussels, Belgium

• The URBIZONE project:– a large meshed outdoor WiFi network

covering the campusses of Brussels universities (ULB-VUB) – 80 antennas

– organized and paid for by the government of the Brussels Region – free for all!

– linked to the IRISnet regional internet (ISP) – which is linked to BELNET

19/5/08 TNC

19/5/08 TNC

Some first conclusive thoughts

URBIZONE is an interesting pilot project– it brings a sponsored network next to the

NREN– it is just an infrastructure, no services –

but that’s NOT the final idea– operators would love to do the same –

with a business model (eg. tourism, health)– the telcos and the media world are today

the clear winners on the Internet – how far should they stay from research networks?

19/5/08 TNC

Final executive summary

• We have far exceeded our wildest expectations/dreams in terms of Quantity

• We are (happily?) often forgetting about Quality

• The non NREN world is overtaking us with Functionality

• Food for thought: does the NREN world need a (new) business model?

19/5/08 TNC

Final souvenir…

P. VAN BINST

High Speed Networking in Europe: Dream, Nightmare, Utopia, or Reality?

Closing address presented at the Symposium on High Speed Networking for Research in Europe, RARE/CEC, Brussels, January 1991.

Computer Networks and ISDN Systems, North‑Holland, 21, June 1991, 4, 353.

19/5/08 TNC

Final souvenir…

P. VAN BINST

High Speed Networking in Europe: Dream, Nightmare, Utopia, or Reality?

Closing address presented at the Symposium on High Speed Networking for Research in Europe, RARE/CEC, Brussels, January 1991.

Computer Networks and ISDN Systems, North‑Holland, 21, June 1991, 4, 353.

Now for the future of networking in Europe, let’s dream and make our Utopia become Reality!