230 likes | 244 Views
Explore the history and challenges of FUNET, the Finnish University and Research Network, and learn about proposed solutions for rising costs and traffic management.
E N D
Development of Funding Models for FUNET Markus Sadeniemi CSC - Center for Scientific Computing Ltd Markus.Sadeniemi@csc.fi tel +358 9 4572711 fax +358 9 4572302
FUNET • Finnish University and Research Network • started in 1984 • international connections • EARN in 1985 • Internet (NORDUnet) in December 1988 • serves • universities (20), polytechnics (25) • research organisations (25), other (10)
CSC - Center for Scientific Computing Ltd • limited company, but • 100% owned by the Ministry of Education • non-profit • three main services • supercomputer centre • expertise in computational science • FUNET services
FUNET, background • originally a joint university project • still not a customer-provider relation • expenses openly shown to FUNET member organisations • yearly discussion of budget/fees with connected organisations
History • universities divided into three categories: big, medium, small • when research institutes and polytechnics joined, more accuracy was needed • eleven categories • based on organisation size = no. of people • staff+students (non-technical divided by 2) • Ministry of Education paid for all development
Budget • FUNET expenses in 1999, MEURinternational connections 3.1FUNET backbone 0.9other 1.2total 5.2
Problem: exponential growth • traffic volume grows 150% / year • international line cost decreases 40% / year=> 2.5 * 0.6 = 1.5=> increase in international connections 50% / year • increase in national costs less • increase in FUNET budget ~ 30% / year
Solution to rising costs • cut the cost increase • restrict traffic growth • technically • administratively • find the money • it still costs < EUR50 / year / computer • wait for a miracle • collapse of line costs
Restricting traffic • administratively • only professional usage, not for hobbies • for privacy reasons enforcing is problematic • technically • technical restrictions for students • no international traffic? no voice/video? • restricted bandwidth for students • restrict bandwidth for all (the easy way!)
Market approach • volume charging • only effective, if you can change the habits of the end user • charging universities is relatively easy • charging departments in universities difficult • charging individual students is impossible both technically and politically
Working group • high-level group appointed by Min of Edu • conclusions • networking is essential • no hard restrictions, money has to be found • charging based on organisation size, connection capacity and traffic volume • endorse technical rationalisations (cache)
Centralised funding? • centralised funding if • networking still in development phase • networking is something to be endorsed • user funding, if • Internet is a commodity item (like POTS) • conclusion • networking still to be endorsed • Min of Edu will pay about 50% centrally
Requirements for charging model • left for CSC to decide together with users • several contradictory requirements for charging • ”fair” in some sense • reflect true costs • not in conflict with market prices • same regardless of geographic location • predictable, easy to budget
Requirements, cont. • simple • understandable to funding bodies • technically simple now and in the future • organisation can decide, how much to buy • should endorse good networking practises • must not prevent • development, new pilot applications • co-operation, free services
Technical possibilities • size of organisation (people, budget,..) • traffic volume • national/international, in/out • number of computers • connection capacity
New charging model • keep the old fee based on organisation size • price increase covered with volume charge • international traffic • direction abroad->university • previous years usage • measured during working hours • (maximum increase 100%)
Benefits • predictable, known in advance • volume affects the price, although slowly • encourages rational usage • discourages the use of the most expensive resource: inbound international line • doesn’t affect (cheap) national traffic • OK to have national pilots • doesn’t affect services given to others
Why not... • connection capacity • (predictable, easy to understand, easy to change) • would affect national traffic • might end in having two networks • one for international connections • bilateral national connections between universities
Why not... • number of computers • technically DNS count • would discourage registering • machines behind firewalls not seen • national traffic • not important • nuisance to count, sometimes difficult