The future of the Internet and the Web

The future of the Internet and the Web Bill St. Arnaud CANARIE Inc – www.canarie.ca Bill.st.arnaud@canarie.ca CANARIE Inc.- Overview • Federal leadership: Concept born in 1990 out of Industry Canada discussions • Founding: Incorporated in 1993 by industry and academia • Funding: From Industry Canada: For networks and research applications from Canadian Heritage, HRDC, Health Canada • Mission: To facilitate development and use of Canada’s advanced communications infrastructure • Primary stakeholders: Universities, provincial research networks, broader research community, colleges, carriers, IT sector, SMEs, broader education sector, broader health sector, provinces Our vision > In the 1960s computing was dominated by central managed large mainframe computers and dumb terminals – Users were charge by CPU minute and memory usage – A lot of research into optimizing CPU and memory usage But starting in the 1970s at our universities and research centers, users could buy their own mini-computer – Researcher could use computer as little or as much as they want with no additional charges – This led to concepts like UNIX, GUI developed ultimately culminating in the PC Telecommunications and the Internet today is like computing in the 1960s – Users are charged by bandwidth consumed and very centralized management We believe the next generation of Internet will be as revolutionary as the PC > > > Future Network Issues 1. Increasing trend to customers own the network rather than buying managed service 2. New network architectures needed for enterprise, industry and critical infrastructure 3. Web 2.0, Mashups, SOA and web services revolution 4. Challenges in the last mile Driver for customer controlled networks > Increasingly more and more organizations are acquiring their own fiber networks – Universities, schools, hospitals, business > Acquiring fiber in the long haul is very expensive to light and obtain – Alternative is to use “dim fiber” –point to point wavelengths – But want flexibility to do configuration and change management as with dark fiber > Increasingly science and business needs dedicated networks for specific applications and disciplines for high data volume grids e.g. SoX and HIPPIA requirements – Want to be able to manipulate the network in the same way they can manipulate the application Condo fiber: Businesses see the light > http://news.com.com/Dark+fiber+Businesses+see+the+light/210 0-1037_3-5557910.html?tag=sas.email > According to Ford if you are spending more than $7000/year on telecom than dark fiber is a better deal > Lighting up fiber used to be technically difficult, but CWDM has made it a no brianer > Next generation CWDM will allow up to 80 Gbps for less than $5k with 10Gbps wavelength NYC Condo build with Lexent Inc CA*net 4 Network New 72 channel x 40 Gbps ROADM 5 x 10 Gbps Optiputer CRC-Cal IT2 Boston Amsterdam 10G 10G San Diego CA*net 4 Network details > CA*net 4 is NOT a single homogenous network > CA*net 4 is made up of many virtual networks (APNs) dedicated to different communities and applications on a common substrate – High energy physics network – Network for government research labs – Network for distributed computer backplane – Virtual networks for network research > One of the parallel networks is a general purpose IP network – This is the only network that carries IPv6 – All other networks only carry IPv4 What is UCLP? > User Controlled LightPaths – a configuration and provisioning tool built around grid technology using web services > User can combine together various links, routers and switches to produce a wide are network that is under their control using BPEL – Articulated Private Network (APN) – Next generation VPN > Uses Service Oriented Architecture (SOA) and so network can be integrated with other web service applications > With SOA instruments, devices, sensors and wireless devices can also be integrated – All under user control and architecture User Controlled Networks in Canada > ATLAS Canada – – – – – Computing & Storage Resource: U of Toronto, TRIUMF, UVic Data Transfer: U of Alberta, Carleton Storage Challenge: CERN / IBM pilot. DB2 Integrated Cluster Environment at Herzberg Institute in Victoria. Storage backup at UVic. > Canadian Virtual Observatory > Neptune Canada – Canarie’s CIIP project. – Integrated instruments, databases, visualization > Canadian Light Source – Canarie’s CIIP project. – Instruments and lightpaths as grid services > WestGrid – SMP at UofA, Cluster at UBC and Storage at SFU. > Eucalyptus – Architectural Collaborative Design > National Engineering Design Network > McGill Neurological Brain Scan Database – In partnership with UCSD and UoBoston > UQAM BioGrid Canada’s role: Huygens probe Titan/Saturn > Real time eVLBI data from Huygens/Cassini satellite with be distributed from Australian receiving dishes across Canada’s CA*net 4 to correlators in Netherlands > First time demonstration of collection and distribution of satellite eVLBI data using advanced networks > Critical for mid-course correction and analyzing descent into Titan atmosphere > Understanding Titan’s atmosphere will give insight into our own climate and atmosphere Canada as a global network hub > Canada through CA*net 4 is becoming a global hub for international research network connectivity > Countries such as Korea, Japan, Taiwan, Ireland, China, etc are acquiring wavelengths across CA*net 4 and routing their traffic through Canada > New milestones in scientific collaboration have been achieved CA*net 5 Plans > Many universities deploying campus fiber networks and extending ROADM into campus – UBC, Carleton, McGill, UoToronto, OttawaU, UdMontreal, Laval > ROADM network allows us to place all active DWDM gear at campus – Core of network is solely optical > We are expecting big demand from individual researchers and departments to purchase their own wavelengths on CA*net 4 > We also planning wavelength swaps with NYSERnet, MiLR, ORANO, NLR and others – We only charge incremental costs for wavelength – $20K per year per end for 10G – $30K - $40K per year per end for 40G > UCLPv2 software now available to allow users to manage their own lightpath topology both on and off campus Internet research still critical > Internet evolution ( or revolution) is far from over > Internet is expanding from an “information service” to a “critical infrastructure” for all aspects of society > Examples: – New building environment control systems are being deployed using distributed web services and Internet – Oil and gas, process control, monitoring critical physical infrastructure will be done over the Internet – New business models based on “mashups”, grids and SOA – Integrative science > Research and education networks critical for promoting “early adopter” developments Mashups and SOA > The goal of SOA and Mashups is to deploy a range of services that encourage others to build applications which manipulate content and services, permitting users to work with the content and services in the manner of their choosing. > Newer-gen apps such as location services Google Mashups, photo sharing (Flickr), social bookmarking (del.icio.us), and community networks (LinkedIn, MySpace) have raised user expectations. > Users expect to be part of the creative flow of content and processes, not just discovers or consumers. > CANARIE is extending this to the network to allow the user to control and manipulate the network Historical Parallels > Before the Web on-line information was only available through a small number of information providers who charged high fees – Compuserve, Dialogic, etc > The Web allowed millions of creators of information to make it easily accessible to all others at very low cost, bypassing the information middleman > SOA and Mashups extend the WEB paradigm to processes, applications and content > SOA and Mashups is about creation of tools and applications (i.e. services) in variety of fields such as eManufacturing, eLearning, eBusiness, eScience, eHealth, etc At there are millions of web sites, there will be millions of Third wave services The Vision: Cabled Ocean Laboratories > Research areas include: – Plate tectonic processes and earthquake dynamics – Dynamic processes of fluid fluxes and gas hydrates in the sea bed – Regional oceanic/climate dynamics and effects on the marine biota – Deep-sea ecosystem dynamics – Engineering and computational research > Delivering power and communications to instruments on and in the seafloor and through the water column VENUS NEPTUNE The Means: NEPTUNE CANARIE Project To adopt and further refine CANARIE to realize a generic web-service enabled telemetry and control system for VENUS and NEPTUNE > Embrace open standards to further innovation, collaboration and knowledge sharing (Open Source, Linux, OASIS) > Migration from proprietary or legacy instrument control schemes (RS-232, RS485, LECIS, SCADA) > Abstract and service-enable system end-points (instruments, PC control stations, compute clusters, data management or analytical applications) > Introduce a loosely-coupled integration fabric which places the burden on the “service bus” rather than the end-points (SOA, ESB, SOAP/XML) > Exploit reliable asynchronous messaging or synchronous messaging as a means for instrument control or data interchange > Research driven process and service choreography (BPEL, BPEL4WS) > Adopt self describing data packets that can be inspected or transformed in transit (XML) > Provide dynamic and ever-evolving experimental design or conditional handling support via content based routing for alerting, workflow, and event management Many Industries Share CANARIE’s Objectives > The “Extended Internet” will connect information systems to physical assets, products, and devices > Complex business processes often need to connect to resources and systems beyond the world of IT and computer users. > They link network endpoints like RFID, telematics, sensor networks, along with the necessary network protocols bandwidth > It also involves the sorting, sifting, and analysis of data gathered by networks of intelligent devices. Data analysis and business intelligence software, linked with device or sensor networks, helps organizations deal with the onslaught of data that such networks will create. “The Seeds Of The Next Big Thing: Sketching The Fourth Wave Of Growth For The Technology Economy” Forester Research Inc, June 2005 What will drive broadband? Applications or Competition? > Once a technology is perceived as having broad utilitarian value, price, as opposed to features or applications drive penetration > Every year the PC has new applications > But the biggest driver for widespread PC in the home is low cost > And what is the driver for low cost? Competition > ITU in their latest broadband rankings also says that “Competition is the primary driver for broadband penetration” Today’s Problem > Telephone companies are in a desperate race to catch up to cableco – They need to build fiber infrastructure in order to offer same capabilities > Cablecos are holding all the aces with DOCSIS 3.0 (100 Mbps-> Gbps ) and cable TV franchises – Although their video franchise may be undermined by MovieLink and SlingBox > To pay for fiber, telecom companies are talking about building a two tiered Internet – A high speed un-congested channel for the telco traffic particularly aimed at carrying video – Charging service and application providers such as Google, MSN, Yahoo etc for premium access > Teleco and cableos could end up squeezing out ASPs and ISPs from providing effective service to their customers Issues facing telcos (and others) > If cable (or other high speed services) already in place, very difficult for a new entrant – even the telephone company > Any network whether wireless or FTTx, overbuilders, or municipal, requires huge capital outlay with a big risk of slow take-up. In addition: – subject to intense competitive and political pressure from incumbents – no obvious consumer demand for “another” network carrying identical services > Need a new business model > Critical leadership role for university research networks How to introduce competition > To date telecom regulators have focused on “facilities based” competition and “open access or unbundling” > Facilities based competition is ultimately the best solution – Has been very successful in the long haul > Open access and unbundling has been the alternative solution for the last mile – But has had poor track record with incumbents – Most incumbents are demanding regulatory relief from unbundling > Mistaken belief by regulators that wireless can compete effectively with fiber as facilities based competition in the last mile > If fiber is a natural monopoly, particularly in last mile suburban areas, then “structural separation” maybe more important than facilities based competition What if the customer owned the last mile? > Increasingly popular strategy at universities, schools, hospitals, businesses, etc > Various companies are specializing in deploying “condominium” fiber to institutions and enterprises – Each institutions purchase one or more fiber pairs in the fiber cable – Customer responsible for lighting their own fiber strands > Fiber cables terminate a carrier neutral colo facilities where customer can cross connect to service provider(s) of their choice > Since customer owns the last mile they are not restricted to a choice of service provider who has deployed last mile infrastructure i.e. telco > ROI is typically 2-3 years versus purchasing managed service Advantages of customer owned last mile > One time cost of $500 for UNLIMITED bandwidth to a neighbourhood node – May be additional small maintenance and cost recovery fee for fiber about $50/year for 10 years > Facilities based competition in the last mile as the customer is in control of the network > Direct connection to content and application providers > Customer negotiates with ISP or carrier of their choice to install ONT or simple media hub at the premises – Customers who only want high speed Internet can purchase media hub for about $200 which includes laser, Gbe transceiver etc Customer owned fibre in the middle mile School board or City Hall Carrier Owned Fiber Application Service Provider Cable head end Telco Central Office Middle mile condominium Fiber with separate strands owned by school and by service providers School See detail Next slide Colo Facility School 802.11b Last mile customer owned condo fiber Business Average Fiber Penetration to 250-500 homes Last mile customer owned fiber Google Neighbourhood Yahoo Colo RBOC Customer owns fiber strand or micro conduit all the way to Neighborhood Node Colo X Colo Facility Splice Box ISP B School Up to 15 km School X 864 strands Business with dual connections to their own strands Condominium Fiber Trunk Neighbourhood Colo and RPON Only the contracted SP provides return signal OXC Aggregator (AOL or RBOC)) Google Yahoo TDM or WDM return Passive Optical Splitter Service provider can be several km away Active laser & optional CWDM at customer premises Neighborhood Colo Node Customer Controlled or Owned Fiber Long term strategy > In discussions with several communities and groups about possible pilots > Intent is NOT to do a national roll out of FTTh > Hopefully a few showcase demos of customer owned FTTh will stimulate RBOCs and municipalities to adopt the concept of customer owned fiber – Especially for telcos who need strong business case to build out fiber > Advantage to incumbent is that they can get faster return on investment if they sell some strands of fiber to customers and regulatory relief from pressure to be “network neutral”