Les Cottrell, UERJ, Rio De Janiero, Feb 16-20, 2004
This workshop was authorized by the International Committee for Future Accelerators (ICFA). There were about 150 attendees, from 21 nationalities, including a large contingent from Brazil. The main topic was the Digital Divide, its extent, how it is changing, what various regions are doing to address it, what is needed and how to address it.
One of the major themes to come out of the conference is that in some cases there is ample fiber available but it is hard (usually expensive) to get access, so most of it is unlit today. For example nine submarine cables connect S & C. America and the Caribbean with 4000Gbps capability of which only 71.2Mbps capacity is available for use (lit). Another theme was that costs per bit can vary dramatically (order of magnitude) from region to region, the higher costs being proportional to increased monopoly power of the telecom industry. Some places such as Australia have been very successful in getting rights (and even in one case installing its own fiber) to fiber and by-passing the telecoms. Typically the problems are often in the last mile, i.e. within the country rather than to the country from outside. In fact in several cases, e.g. Pakistan and S. America communications between regions may go via London (in the case of Pakistan) or the US (S. America).
The requirements for data intensive science are already present for HENP, Astronomy (e.g. SDSS and LIGO), medicine (MammoGrid Project, 1 mammogram ~ 10MBytes, 28PBytes accumulated / year just in US hospitals, using many technologies developed by HENP), 1 yr LHC data is 20PB ~ 20km CD stacked up. To address such needs a concept of computational grids is being embraced to enable shared cpu power and data storage.
UERJ (the host university of the meeting) upgraded from 20Mbps to 34Mbps, Feb 14, 2004.. UERJ wants to move from being an HENP Tier2 to Tier 1 site, the university of Sao Paolo wants to move from tier 3 to 2. A Gbps ring is being built for Rio and Sao Paolo. 12% of Brazilians have access to PC, and only 8% to Internet.
UERJ recognizes need to improve networks to enable grid capabilities.
"World Telecomm Development Report 2003 - Access Indicators for Information Society" (256 papers). 1988 8 countries connected to Internet, now about 209. Only 3% of Egyptians have access to Internet, but have 2nd lowest price in world.
Hard to convince authorities and opinion makers that it is absolutely necessary to upgrade digital networks for science. This is the century of Science and Technology. The tools for discoveries were and are being built to allow us to progress on our knowledge of nature. The Digital Divide is a major enemy.
Digital Divide is deepening. Performance, in particular in backbones, is improving (last 2 years factor ~4 to several hundred times). Bottlenecks may be technical or political, often in last mile.
MDS=Grid Information Services (finds what resources are available). There are multiple grids being formed, e.g. PPDG. iVDGL+, based on open standards, would like adoption by industry to provide tools. Standard architecture/substrate is OGSA (Grid services) based on standard web services. Web services provide discovery (registry) and access for services. The need for standards has reduced the flexibility for individual contributions. OGSI has moved to be more web service compatible.
Started discussion in 1985. Now in process to make decision for construction. Have gone through 2 designs. Design was a network effort involving 6 participant countries including Japan, Russia, Europe and US. Have an electronic model of ITER. Will use GLORIAD for networking. There is a big effort to enable remote participation for collaboration and remote control.
GLORIAD is currently OC3. Hope to move GLORIAD to 10Gbps optical wavelength network for N. Hemisphere (Russia, China, US). One goal is to reduce the Digital Divide for Russia & China. Lead Russian institute is Kurchatov Institute.
The Role of Science in the Information Society (RSIS) part of the World Summit on the Information Socitey (WSIS) held in Geneva in December. There were documents produced (declaration of principle and an action plan), question is how valuable the documents will be. Funding is unclear. Ca;l for action: connect all universities, poly-tech high schools by creating a Global Research and Education Network by 2005; create basic, affordable (open source) software tools for disseminating information; create open Courseware & digital certified knowledge repositories.
Cost of fiber ~ 1.3 cost of equipment to light. Cost of optical equipment 10% switching, 10% of full routing equipment for same throughput: 100B pkt at 40GBps == 20ns to look up in 140kEntries routing table so look for hybrid architectures which serve all classes and distances in a cost effective way. Class C users point to point few users high bandwidth do not need routing so use sub-lambda, ethernet-sdh, class A & use routers for world. Separate long term high performance flows to use lambda switching.
Latin America communicates internally via Miami, also communicates with the rest of the world. Compares with Europe in the 80's and early 90's when Europe communicated internally vai the US. Now Europe has gone through TEN-34, TEN-155 GEANT. CLARA being founded to provide a Pan Latin American network being funded partially by Europe (in particular leadership coming from Portugal and Spain. Expect to start operations in May this year. A ring to connect major 4 countries with other countries connecting to 4 countries (MX-BR-AR-CL-PA) at 10 and 45Mbps. 155Mbps ring, Caribbean connects via Panama. Traditional costs to connect up within Latin America were like $10M for 34Mbps. Will be R&E only, will connect to Europe at 622Mbps, might be a cross-border connection in say Tijuana to Abilene.
Brazil/Sao Paolo/UERJ to monitor within L. America.
Australia will have 3 international circuits: OC3 Sidney-Fiji-Hillsboro; have 622Mbps to PAIX and coming soon to LA; Southern Cross dual 10Gbps (STM-64) experimental via Hawaii to Seattle to be activated mid 2004, looking to add Mauna Kea (for telescope). Possibly connect NZ later. These international circuits managed by AARNET (supports A&R nets in Australia, it is a Not For Profit owned by universities.) In 2001 had 155Mbps to US. Developed relationships with power companies, gas pipelines (to Tasmania) and construction companies to get Gbps capacity to regional areas where dealing with traditional telcos was not feasible. Have framework using satellite to connect up Pacific Islands. AARnet has a carrier license and has actually once carried out civil works to lay fiber. Two fiber rollouts is Australia recently went into receivership, one was very interesting for AARnet clients (Universities, research and astronomers), AARnet got access and partner with construction company. Have 2 fibre pairs Australia wide. Have access to environment vaults. Using Procket routers. Switches are Cisco 6509s. They have dual PoPs in the major cities.
The Cairns APAN meeting will include a CCIRN side-meeting which addresses inter-continental concerns and is of interest to our community.
I met with Arshad Ali and discussed collaborations between NIIT in Rawalpindi and SLAC to work on network monitoring and to collaborate on the MAGGIE proposal SLAC, LBNL, Internet2, U Delaware, and NCSA are submitting to the DoE/MICS office.
I met with Dave Foster of CERN to discuss extending the deployment of IEPM-BW to CERN.
Vicky White of FNAL and I discussed ongoing funding for PingER, she suggests visiting NSF in Washington with a proposal for support for PingER for the Digital Divide/hard to reach places.
Sergio Novaes of AMPATH/FIU agreed to provide contacts for monitoring in Latin America. Jorge Valerio of UERJ <firstname.lastname@example.org> said he would provide assistance to get an account at UERJ in Rio de Janeiro and to set up a PingER monitoring host.
I met with Alexei Soldatov <email@example.com> of the Kurchatov Institute and he agreed to provide assistance for hostrs to monitor in Russia.