High-Speed Internet Switches and Routers COMP 680E Mounir Hamdi Professor, Computer Science Director, MSc-IT Hong Kong University of Science and Technology

Goals of the Course Understand the architecture, operation, and evolution of the Internet IP, ATM, Optical Understand how to design, implement and evaluate Internet routers and switches (Telecom Equipment) Both hardware and software solutions Get familiar with current Internet switches/routers research and development efforts Appreciate what is a good project Task selection and aim Survey & solution & research methodology Presentation Apply what you learned in a small class project

Outline of the Course The focus of the course is on the design and analysis of high-performance electronic/optical switches/routers needed to support the development and delivery of advanced network services over high-speed Internet. The switches and routers are the KEY building blocks of the Internet, and as a result, the capability of the Internet in all its aspects depends on the capability of its switches and routers. The goal of the course is to provide a basis for understanding, appreciating, and performing research and development in networking with a special emphasis on switches and routers.

Outline of the Course Introduction Definition and History of Networking/Internet Evolution and Trends in the Internet Architecture of The Internet Classification and Evolution of Internet Equipment Review and Evolution of Internet Protocols Different technologies of the Internet

Outline of the Course Network Processors: Table Lookup and Packet Classification Internet addressing and CIDR Table Lookup: Exact matches, longest prefix matches, performance metrics, hardware and software solutions. Packet classifiers for firewalls, QoS, and policy-based routing; graphical description and examples of 2-D classification, examples of classifiers, theoretical and practical considerations State-of-the-art commercial products

Outline of the Course High-Performance Packet Switches/Routers Architectures of packet switches/routers (IQ, OQ, VOQ, CIOQ, SM, Buffered Crossbars) Design and analysis of switch fabrics (Crossbar, Clos, shared memory, etc.) Design and analysis of scheduling algorithms (arbitration, Maximum/maximal matching, shared memory contention, etc.) Emulation of output-queueing switches by more practical switches State-of-the-art commercial products

Outline of the Course Quality-of-Service Provision in the Internet QoS paradigms (IntServ, DiffServ, Controlled load, etc.) MPLS/GMPLS Flow-based QoS frameworks: Hardware and software solutions Stateless QoS frameworks: RED, WRED, congestion control, and Active queue management State-of-the-art commercial products

Outline of the Course Optical Networks Optical technology used for the design of switches/routers as well as transmission links Dense Wavelength Division Multiplexing Optical Circuit Switches: Architectural alternatives and performance evaluation Optical Burst switches Optical Packet Switches Design, management, and operation of DWDM networks State-of-the-art commercial products

Grading Homework 20% Midterm 30% Project 50%

Course project Investigate existing advances and/or new ideas and solutions – related to Internet Switches and Routers - in a small scale project (To be given or chosen on your own) define the problem execute the survey and/or research work with your partner write up and present your finding

Course Project I’ll post on the class web page a list of projects you can either choose one of these projects or come up with your own Choose your project, partner (s), and submit a one page proposal describing: the problem you are investigating your plan of project with milestones and dates any special resources you may need Final project presentation (~ 30 minutes) Submit project papers

Homework Goals: Synthesize main ideas and concepts from very important research or development work I will post in the class web page a list of “well-known” papers to choose from Report contains: Description of the papers Goals and problems solved in the papers What did you like/dislike about the paper Recommendations for improvements or extension of the work

How to Contact Me Instructor: Mounir Hamdi hamdi@cs.ust.hk Office Hours You can come any time – just email me ahead of time I would like to work closely with each student

Overview and History of the Internet

What is a Communication Network? (from an end system point of view) A network offers a service: move information Messenger, telegraph, telephone, Internet … another example, transportation service: move objects horse, train, truck, airplane ... What distinguishes different types of networks? The services they provide What distinguish the services? latency bandwidth loss rate number of end systems Reliability, unicast vs. multicast, real-time, message vs. byte ...

What is a Communication Network? Infrastructure Centric View Hardware Electrons and photons as communication data Links: fiber, copper, satellite, … Switches: mechanical/electronic/optical, Software Protocols: TCP/IP, ATM, MPLS, SONET, Ethernet, PPP, X.25, Frame Relay, AppleTalk, IPX, SNA Functionalities: routing, error control, congestion control, Quality of Service (QoS), … Applications: FTP, WEB, X windows, VOIP, IPTV...

Types of Networks Geographical distance Information type Personal Areas Networks (PAN) Local Area Networks (LAN): Ethernet, Token ring, FDDI Metropolitan Area Networks (MAN): DQDB, SMDS (Switched Multi-gigabit Data Service) Wide Area Networks (WAN): IP, ATM, Frame relay Information type data networks vs. telecommunication networks Application type special purpose networks: airline reservation network, banking network, credit card network, telephony general purpose network: Internet

Types of Networks Right to use Ownership of protocols Technologies private: enterprise networks public: telephony network, Internet Ownership of protocols proprietary: SNA open: IP Technologies terrestrial vs. satellite wired vs. wireless Protocols IP, AppleTalk, SNA

The Internet Global scale, general purpose, heterogeneous-technologies, public, computer network Internet Protocol Open standard: Internet Engineering Task Force (IETF) as standard body Technical basis for other types of networks Intranet: enterprise IP network Developed by the research community

1961-1972: Early packet-switching principles Internet History 1961-1972: Early packet-switching principles 1961: Kleinrock - queueing theory shows effectiveness of packet-switching 1964: Baran – Introduced first Distributed packet-switching Communication networks 1967: ARPAnet conceived and sponsored by Advanced Research Projects Agency – Larry Roberts 1969: first ARPAnet node operational at UCLA. Then Stanford, Utah, and UCSB 1972: ARPAnet demonstrated publicly NCP (Network Control Protocol) first host-host protocol (equivalent to TCP/IP) First e-mail program to operate across networks ARPAnet has 15 nodes and connected 26 hosts

1972-1980: Internetworking, new and proprietary nets Internet History 1972-1980: Internetworking, new and proprietary nets 1970: ALOHAnet satellite network in Hawaii 1973: Metcalfe’s PhD thesis proposes Ethernet 1974: Cerf and Kahn - architecture for interconnecting networks (TCP) late70’s: proprietary architectures: DECnet, SNA, XNA late 70’s: switching fixed length packets (ATM precursor) 1979: ARPAnet has 200 nodes Cerf and Kahn’s internetworking principles: minimalism, autonomy - no internal changes is required to interconnect networks best effort service model stateless routers decentralized control define today’s Internet architecture

1971-1973: Arpanet Growing 1970 - First 2 cross-country link, UCLA-BBN and MIT-Utah, installed by AT&T at 56kbps Initial ARPAnet was a single closed network – to communicate with an ARPA host one had to be attached to another ARPAnet IMP

1980-1990: new protocols, a proliferation of networks Internet History 1980-1990: new protocols, a proliferation of networks 1983: deployment of TCP/IP 1982: SMTP e-mail protocol defined 1983: DNS defined for name-to-IP-address translation 1985: ftp protocol defined (first version: 1972) 1988: TCP congestion control New national networks: CSnet, BITnet, NSFnet, Minitel 100,000 hosts connected to confederation of networks

1990’s: commercialization, the WWW Internet History 1990’s: commercialization, the WWW Early 1990’s: ARPAnet decomissioned 1991: NSF lifts restrictions on commercial use of NSFnet (decommissioned, 1995) early 1990s: WWW hypertext [Bush 1945, Nelson 1960’s] HTML, http: Berners-Lee 1994: Mosaic, later Netscape late 1990’s: commercialization of the WWW Late 1990’s: est. 50 million computers on Internet est. 100 million+ users in 160 countries backbone links running at 1 Gbps+ 2000’s VoIP, Video on demand, Internet business RSS, Web 2.0

Growth of the Internet Number of Hosts on the Internet: Aug. 1981 213 Oct. 1984 1,024 Dec. 1987 28,174 Oct. 1990 313,000 Oct. 1993 2,056,000 Apr. 1995 5,706,000 Jan. 1997 16,146,000 Jan. 1999 56,218,000 Jan. 2001 109,374,000 Jan. 2003 171,638,297 Jul 2004 285,139,107 Jul 2005 353,284,187 Today ~ 440,000,000 Source: http://www.isc.org/index.pl?/ops/ds/host-count-history.php

Internet - Global Statistics 2005 350 Million Hosts 1,018 Million Users 1997 22.5 Million Hosts 50 Million Users (approx. 2.4 Billion Telephone Terminations, 660 Million PCs and 1.6B mobile phones) 4

Internet Penetration December 2006 (Source www.internetstats.com)

Top 10: % Internet Use (Dec 2006) Country or Region Penetration (% Population) % Internet Users 1 Iceland 86.3 % 2 New Zealand 74.9 % 3 Sweden 74.7 % 4 Portugal 73.8 % 5 Australia 70.2 % 6 United States 69.6 % 7 Falkland Islands 69.4 % 8 Denmark 69.2 % 9 Hong Kong (China) 68.2 % 10 Luxembourgh 68.0 % www.internetworldstats.com

Languages of Internet Users

Who is Who on the Internet ? Internet Engineering Task Force (IETF): The IETF is the protocol engineering and development arm of the Internet. Subdivided into many working groups, which specify Request For Comments or RFCs. IRTF (Internet Research Task Force): The Internet Research Task Force is composed of a number of focused, long-term and small Research Groups. Internet Architecture Board (IAB): The IAB is responsible for defining the overall architecture of the Internet, providing guidance and broad direction to the IETF. The Internet Engineering Steering Group (IESG): The IESG is responsible for technical management of IETF activities and the Internet standards process. Composed of the Area Directors of the IETF working groups.

Internet Standardization Process All standards of the Internet are published as RFC (Request for Comments). But not all RFCs are Internet Standards ! available: http://www.ietf.org A typical (but not only) way of standardization is: Internet Drafts RFC Proposed Standard Draft Standard (requires 2 working implementation) Internet Standard (declared by IAB) David Clark, MIT, 1992: "We reject: kings, presidents, and voting. We believe in: rough consensus and running code.”

Services Provided by the Internet Shared access to computing resources telnet (1970’s) Shared access to data/files FTP, NFS, AFS (1980’s) Communication medium over which people interact email (1980’s), on-line chat rooms, instant messaging (1990’s) audio, video (1990’s) replacing telephone network? A medium for information dissemination USENET (1980’s) WWW (1990’s) replacing newspaper, magazine? replacing radio, CD, TV? 2000s: peer-to-peer systems – triple play bundles

Today’s Vision Everything is digital: voice, video, music, pictures, live events, … Everything is on-line: bank statement, medical record, books, airline schedule, weather, highway traffic, … Everyone is connected: doctor, teacher, broker, mother, son, friends, enemies

What is Next? – many of it already here Electronic commerce virtual enterprise Internet entertainment interactive sitcom World as a small village community organized according to interests enhanced understanding among diverse groups Electronic democracy little people can voice their opinions to the whole world little people can coordinate their actions bridge the gap between information haves and have no’s Electronic Crimes hacker can bring the whole world to its knee

Industrial Players Telephone companies own long-haul and access communication links, customers Cable companies own access links Wireless/Satellite companies alternative communication links Utility companies: power, water, railway own right of way to lay down more wires Medium companies own content Internet Service Providers Equipment companies switches/routers, chips, optics, computers Software companies

What is the Internet? The collection of hosts and routers that are mutually reachable at any given instant All run the Internet Protocol (IP) Version 4 (IPv4) is the dominant protocol Version 6 (IPv6) is the future protocol Lots of protocols below and above IP, but only one IP Common layer

Commercial Internet after 1994 Roughly hierarchical National/international backbone providers (NBPs) e.g., Sprint, AT&T, UUNet interconnect (peer) with each other privately, or at public Network Access Point (NAPs) regional ISPs connect into NBPs local ISP, company connect into regional ISPs NBP A NBP B NAP regional ISP local ISP

Internet Organization NAP BSP ISP POP CN ISP = Internet Service Provider BSP = Backbone Service Provider NAP = Network Access Point POP = Point of Presence CN = Customer Network

Commercial Internet after 1994 Joe's Company Stanford Berkeley Campus Network Regional ISP Bartnet Xerox Parc SprintNet America On Line UUnet NSF Network IBM NSF Network Modem Internet MCI IBM

Internet Architecture

Basic Architecture: NAPs and National ISPs The Internet has a hierarchical structure. At the highest level are large national Internet Service Providers that interconnect through Network Access Points (NAPs). There are about a dozen NAPs in the U.S., run by common carriers such as Sprint and Ameritech, and many more around the world (Many of these are traditional telephone companies, others are pure data network companies).

The real story… Regional ISPs interconnect with national ISPs and provide services to their customers and sell access to local ISPs who, in turn, sell access to individuals and companies.

pop pop pop pop

The Hierarchical Nature of the Internet Metro Network Long Distance Network Central Office Central Office Node Node San Francisco New York Central Office Major City - Regional Center Major City - Regional Center Central Office Telephone instrument is dumb … lo tech Telephone network is incredibly complex and in flexible Grew over decades, impossible to upgrade, monumental investment Central Office Central Office Node Node

Points of Presence (POPs) A B C POP1 POP3 POP2 POP4 D E F POP5 POP6 POP7 POP8

A Bird’s View of the Internet

A Bird’s View of the Internet

Hop-by-Hop Behavior Within HK Los Angeles Qwest (Backbone) Stanford From traceroute.pacific.net.hk to cs.stanford.edu traceroute to cs.stanford.edu (171.64.64.64) from lamtin.pacific.net.hk (202.14.67.228), rsm-vl1.pacific.net.hk (202.14.67.5) gw2.hk.super.net (202.14.67.2) 3 wtcr7002.pacific.net.hk (202.64.22.254) 4 atm3-0-33.hsipaccess2.hkg1.net.reach.com (210.57.26.1) 5 ge-0-3-0.mpls1.hkg1.net.reach.com (210.57.2.129) 6 so-4-2-0.tap2.LosAngeles1.net.reach.com (210.57.0.249) 7 unknown.Level3.net (209.0.227.42) 8 lax-core-01.inet.qwest.net (205.171.19.37) 9 sjo-core-03.inet.qwest.net (205.171.5.155) 10 sjo-core-01.inet.qwest.net (205.171.22.10) 11 svl-core-01.inet.qwest.net (205.171.5.97) 12 svl-edge-09.inet.qwest.net (205.171.14.94) 13 65.113.32.210 (65.113.32.210) 14 sunet-gateway.Stanford.EDU (171.66.1.13) 15 CS.Stanford.EDU (171.64.64.64) Within HK Los Angeles Qwest (Backbone) Stanford

NAP-Based Architecture UUNET NY NAP CHI WDC SF MCI QWest Sprint Net MAE West

Basic Architecture: MAEs and local ISPs As the number of ISPs has grown, a new type of network access point, called a metropolitan area exchange (MAE) has arisen. There are about 50 such MAEs around the U.S. today. Sometimes large regional and local ISPs (AOL) also have access directly to NAPs. It has to be approved by the other networks already connected to the NAPs – generally it is a business decision.

Internet Packet Exchange Charges Peering ISPs at the same level usually do not charge each other for exchanging messages. They update their routing tables with each other customers or pop. This is called peering.

Charges: Non-Peering Higher level ISPs, however, charge lower level ones (national ISPs charge regional ISPs which in turn charge local ISPs) for carrying Internet traffic. Local ISPs, of course, charge individuals and corporate users for access.

Connecting to an ISP ISPs provide access to the Internet through a Point of Presence (POP). Individual users access the POP through a dial-up line using the PPP protocol. The call connects the user to the ISP’s modem pool, after which a remote access server (RAS) checks the userid and password.

More on connecting Once logged in, the user can send TCP/IP/[PPP] packets over the telephone line which are then sent out over the Internet through the ISP’s POP (point of presence) Corporate users might access the POP using a T-1, T-3 or ATM OC-3 connections, for example, provided by a common carrier.

DS (telephone carrier) Data Rates Designation Number of Voice Circuits Bandwidth DS0 1 64 kb/s DS1 (T1) 24 1.544 Mb/s DS2 (T2) 96 6.312 Mb/s DS3 (T3) 672 44.736 Mb/s

Optical Level Line Rate, Mb/s SONET Data Rates A small set of fixed data transmission rates is defined for SONET. All of these rates are multiples of 51.84 Mb/s, which is referred to as Optical Carrier Level 1 (on the fiber) or Synchronous Transport Signal Level 1 (when converted to electrical signals) Optical Level Line Rate, Mb/s OC-1 OC-3 OC-9 OC-12 OC-18 OC-24 OC-36 OC-48 OC-96 OC-192 OC-768 51.840 155.520 466.560 622.080 933.120 1244.160 1866.240 2488.320 4976.640 9953.280 39813.120

ISPs and Backbones POP: connection with POP of the same ISP or different ISPs POP: Connection with customers T1 Lines to Customers T3 Lines to Other POPs Line Server Dialup Lines to Customers T3 Line OC-3 Line ATM Switch Router Core Router Ethernet OC-3 Lines to Other ATM Switches Point of Presence (POP)

ISP Point-of-Presence Individual Dial-up Customers ISP POP Modem Pool ISP POP Corporate T1 Customer T1 CSU/DSU ATM Switch ATM Switch Corporate T3 Customer ISP POP T3 CSU/DSU Remote Access Server Corporate OC-3 Customer ATM Switch NAP/MAE

HK Major Internet Exchange (HK –NAP/ MAE)

From the ISP to the NAP/MAE Each ISP acts as an autonomous system, with is own interior and exterior routing protocols. Messages destined for locations within the same ISP are routed through the ISP’s own network. Since most messages are destined for other networks, they are sent to the nearest MAE or NAP where they get routed to the appropriate “next hop” network.

From the ISP to the NAP/MAE Next is the connection from the local ISP to the NAP. From there packets are routed to the next higher level of ISP. Actual connections can be complex and packets sometimes travel long distances. Each local ISP might connect a different regional ISP, causing packets to flow between cities, even though their destination is to another local ISP within the same city.

Inside an Internet Network Access Point ISP A ISP D Router Router ATM Switch ISP B ISP E Router ATM Switch ISP C Route Server ISP F Router ATM Switch

Inside an Internet Network Access Point

Network Access Point

ISPs and Backbones ATM/SONET Core Router Core Access Network POP POP

Three national ISPs in North America

Backbone Map of UUNET - USA

UUNET Mixed OC-12 – OC-48 – OC 192 backbone 1000s miles of fiber 3000 POPs 2,000,000 dial-in ports

Backbone Map of UUNET - World

Qwest OC-192 backbone 25,000 miles of fiber 635 POPs 85,000 dial-in ports

AT&T OC-192 backbone 53,000 miles of fiber 2000 POPs 0 dial-in ports

Internet Backbones in 2006 As of mid-2001, most backbone circuits for national ISPs in the US are 622 Mbps ATM OC-12 lines. The largest national ISPs are planning to convert to OC-192 (10 Gbps) by the end of 2003. A few are now experimenting with OC-768 (40 Gbps) and some are planning to use OC-3072 (160 Gbps). Aggregate Internet traffic reached 2.5 Terabits per second (Tbps) by mid-2001. It is expected to reach 35 Tbps by 2007.

Links for Long Haul Transmission Possibilities IP over SONET IP over ATM IP over Frame Relay IP over WDM

User Services & Core Transport EDGE CORE OC-3 OC-12 STS-1 Frame Relay Frame Relay IP Router IP ATM Switch ATM Sonet ADM Lease Lines TDM Switch Users Services Service Provider Networks Transport Provider Networks

Typical (BUT NOT ALL) IP Backbone (Late 1990’s) SONET/SDH DCS Core Router ATM Switch MUX ADM This slide shows equipment layering for a typical IP backbone in the late 1990’s. Data was piggybacked over a traditional voice/TDM transport network. Historically, this made sense. Today, it doesn’t. Data piggybacked over traditional voice/TDM transport

IP Backbone Evolution (One version) Core Router (IP/MPLS) SONET/ SDH DWDM Core Router (IP/MPLS) Removal of ATM Layer Next generation routers provide trunk speeds and SONET interfaces Multi-protocol Label Switching (MPLS) on routers provides traffic engineering FR/ATM Switch MUX If speed matching is no longer an issue, then we can remove the mux. SONET/SDH DWDM (Maybe)

Hierarchy of Routers and Switches Core IP Router FR/ATM Switch SONET/SDH IP Router (datagram packet switching) Deals directly with IP addresses; Slow – typically no interface to SONET equipment Expensive Efficient (No header overhead and alternative routing) ATM Switch (VC packet switching) Label based switching Fast (Hardware forwarding) Header Tax SONET OXC (Circuit switching) Extremely fast – Optical technology Inexpensive If speed matching is no longer an issue, then we can remove the mux.

Customer Network All hosts owned by a single enterprise or business Common case Lots of PCs Some servers Routers Ethernet 10/100/1000-Mb/s LAN T1/T3 1.54/45-Mb/s wide area network (WAN) connection

Customer Network Clients LAN Servers Router WAN Ethernet 10 Mb/s T1 Link 1.54 Mb/s

Internet Access Technologies

Internet Access Technologies Previously, most people use 56K dial-up lines to access the Internet, but a number of new access technologies are now being offered. The main new access technologies are: Digital Subscriber Line/ADSL Cable Modems Fixed Wireless (including satellite access) Mobile Wireless (WAP)

Digital Subscriber Line Digital Subscriber Line (DSL) is one of the most used technologies now being implemented to significantly increase the data rates over traditional telephone lines. Historically, voice telephone circuits have had only a limited capacity for data communications because they were constrained by the 4 kHz bandwidth voice channel. Most local loop telephone lines actually have a much higher bandwidth and can therefore carry data at much higher rates.

Digital Subscriber Line DSL services are relatively new and not all common carriers offer them. Two general categories of DSL services have emerged in the marketplace. Symmetric DSL (SDSL) provides the same transmission rates (up to 128 Kbps) in both directions on the circuits. Asymmetric DSL (ADSL) provides different data rates to (up to 640 Kbps) and from (up to 6.144 Mbps) the carrier’s end office. It also includes an analog channel for voice transmissions.

DSL Architecture Customer Premises Local Carrier End Office DSL Modem Line Splitter Main Distribution Frame Voice Telephone Network Local Loop Hub Telephone ISP POP ATM Switch Computer DSL Access Multiplexer Computer ISP POP Customer Premises ISP POP ISP POP Customer Premises

Cable Modems One potential competitor to DSL is the “cable modem” a digital service offered by cable television companies which offers an upstream rate of 1.5-10 Mbps and a downstream rate of 2-30 Mbps. A few cable companies offer downstream services only, with upstream communications using regular telephone lines.

Cable Modem Architecture Customer Premises Cable Company Fiber Node Cable Company Distribution Hub TV Video Network Cable Modem Cable Splitter Downstream Combiner Optical/Electrical Converter Upstream Hub TV Router Shared Coax Cable System Cable Company Fiber Node Cable Modem Termination System Computer Computer ISP POP Customer Premises Customer Premises Cable Modem Architecture

Fixed Wireless Fixed Wireless is another “dish-based” microwave transmission technology. It requires “line of sight” access between transmitters. Data access speeds range from 1.5 to 11 Mbps depending on the vendor. Transmissions travel between transceivers at the customer premises and ISP’s wireless access office.

Fixed Wireless Architecture Customer Premises Individual Premise Main Distribution Frame Voice Telephone Network DSL Modem Line Splitter Hub Individual Premise Telephone Wireless Transceiver DSL Access Multiplexer Individual Premise Computer Computer Wireless Access Office Customer Premises Wireless Transceiver Router Customer Premises ISP POP

Classifying Computer Networks Please if anyone has additional comments please speak up

A Taxonomy of Communication Networks Communication networks can be classified based on the way in which the nodes exchange information: Communication Network Switched Communication Network Broadcast Communication Network Circuit-Switched Communication Network Packet-Switched Communication Network Datagram Network Virtual Circuit Network

Broadcast vs. Switched Communication Networks Broadcast communication networks information transmitted by any node is received by every other node in the network examples: usually in LANs (Ethernet, Wavelan) Problem: coordinate the access of all nodes to the shared communication medium (Multiple Access Problem) Switched communication networks information is transmitted to a sub-set of designated nodes examples: WANs (Telephony Network, Internet) Problem: how to forward information to intended node(s) this is done by special nodes (e.g., routers, switches) running routing protocols

Circuit Switching Three phases circuit establishment data transfer circuit termination If circuit is not available: “Busy signal” Examples Telephone networks ISDN (Integrated Services Digital Networks) Optical Backbone Internet (going in this direction)

Timing in Circuit Switching Host 1 Host 2 Node 1 Node 2 DATA processing delay at Node 1 propagation delay between Host 1 and Node 1 Circuit Establishment Data Transmission Circuit Termination propagation delay between Host 2 and Node 1

Circuit Switching A node (switch) in a circuit switching network incoming links Node outgoing links

Circuit Switching: Multiplexing/Demultiplexing Time divided in frames and frames divided in slots Relative slot position inside a frame determines which conversation the data belongs to If a slot is not used, it is wasted There is no statistical gain

Packet Switching Data are sent as formatted bit-sequences, so-called packets. Packets have the following structure: Header and Trailer carry control information (e.g., destination address, check sum) Each packet is passed through the network from node to node along some path (Routing) At each node the entire packet is received, stored briefly, and then forwarded to the next node (Store-and-Forward Networks) Typically no capacity is allocated for packets Header Data Trailer

Packet Switching A node in a packet switching network incoming links outgoing links Memory

Packet Switching: Multiplexing/Demultiplexing Data from any conversation can be transmitted at any given time How to tell them apart? use meta-data (header) to describe data

Datagram Packet Switching Each packet is independently switched each packet header contains destination address No resources are pre-allocated (reserved) in advance Example: IP networks

Timing of Datagram Packet Switching Host 1 Host 2 Node 1 Node 2 propagation delay between Host 1 and Node 2 transmission time of Packet 1 at Host 1 Packet 1 Packet 2 Packet 3 processing delay of Packet 1 at Node 2 Packet 1 Packet 2 Packet 3 Packet 1 Packet 2 Packet 3

Datagram Packet Switching Host C Host D Host A Node 1 Node 2 Node 3 Node 5 Host B Host E Node 7 Node 6 Node 4

Virtual-Circuit Packet Switching Hybrid of circuit switching and packet switching data is transmitted as packets all packets from one packet stream are sent along a pre-established path (=virtual circuit) Guarantees in-sequence delivery of packets However: Packets from different virtual circuits may be interleaved Example: ATM networks

Virtual-Circuit Packet Switching Communication using virtual circuits takes place in three phases VC establishment data transfer VC disconnect Note: packet headers don’t need to contain the full destination address of the packet (One key to this idea)

Timing of VC Packet Switching Host 1 Host 2 Node 1 Node 2 propagation delay between Host 1 and Node 1 VC establishment Packet 1 Packet 2 Packet 3 Packet 1 Packet 2 Packet 3 Data transfer Packet 1 Packet 2 Packet 3 VC termination

VC Packet Switching Host C Host D Host A Node 1 Node 2 Node 3 Node 5 Host B Host E Node 7 Node 6 Node 4

Packet-Switching vs. Circuit-Switching Most important advantage of packet-switching over circuit switching: Ability to exploit statistical multiplexing: efficient bandwidth usage; ratio between peek and average rate is 3:1 for audio, and 15:1 for data traffic However, packet-switching needs to deal with congestion: more complex routers harder to provide good network services (e.g., delay and bandwidth guarantees) In practice they are combined IP over SONET, IP over Frame Relay

Fixed-Rate versus Bursty Data

Packet Switches Connectionless Packet Switch Connection-Oriented Routing Table Destination Address Connectionless Packet Switch A A Possibly different paths through switch A Connection Identifier B B Always same path through switch B Connection-Oriented Packet Switch Connec- tion Table

Store-and-Forward Operation Packet entering switch or router is stored in a queue until it can be forwarded Queueing Header processing Routing-table lookup of destination address Forwarding to next hop Queueing time variation can result in non-deterministic delay behavior (maximum delay and delay jitter) Packets might overflow finite buffers (Network congestion)

Link Diversity Internet meant to accommodate many different link technologies Ethernet ATM SONET ISDN Modem The list continues to grow “IP on Everything”

Internet Protocols

Internet Protocols Host Router Host Application Application Transport Network Network Network Link Link Link Link Host Router Host

IP Protocol Stack Ping Telnet FTP H.323 SIP RTSP RSVP S/MGCP/ NCS User application TCP UDP OSPF ARP ICMP IP IGMP RARP Link Layer

Demultiplexing Transport Network Link incoming frame RARP ARP UDP TCP Application TCP IGMP ICMP Ethernet Driver IP Transport Network Link

Link Protocols Numerous link protocols Ethernet + LLC (Logical Link Control) T1/DS1 + HDLC (High-level Data Link Control) T3/DS3 + HDLC Dialup + PPP (Point-to-Point Protocol) ATM/SONET + AAL (ATM Adaptation Layer) ISDN + LAPD (Link Access Protocol) + PPP FDDI + LLC

Additional Link Protocols ARP (Address Resolution Protocol) is a protocol for mapping an IP address to a physical machine address that is recognized in the local network. Most commonly, this is used to associate IP addresses (32-bits long) with Ethernet MAC addresses (48-bits long). RARP is the reverse of ARP

ARP Protocol

Sending an IP Packet over a LAN

Transport Protocols Transmission Control Protocol (TCP) User Datagram Protocol (UDP)

Application Protocols File Transfer Protocol (FTP) Simple Mail Transfer Protocol (SMTP) Telnet Hypertext Transfer Protocol (HTTP) Simple Network Management Protocol (SNMP) Remote Procedure Call (RPC) DNS: The Domain Name System service provides TCP/IP host name to IP address resolution.

The Internet Network layer: The Glue of all Networks Transport layer: TCP, UDP IP protocol addressing conventions datagram format packet handling conventions Routing protocols path selection RIP, OSPF, BGP Network layer routing table ICMP protocol error reporting router “signaling” Link layer physical layer

Demultiplexing Details echo server 1024-5000 7 FTP server telnet server 21 23 discard server User process 9   TCP src port TCP dest port data header UDP 17 TCP ICMP 1 TCP 6 IGMP 2  dest addr source data protocol type IP header hdr cksum ARP x0806 Others x8035 IP RARP Novell IP x0800 AppleTalk  dest addr source addr Ethernet frame type data CRC (Ethernet frame types in hex, others in decimal)

IP Features Connectionless service Addressing Data forwarding Fragmentation and reassembly Supports variable size datagrams Best-effort delivery: Delay, out-of-order, corruption, and loss possible. Higher layers should handle these. Provides only “Send” and “Delivery” services Error and control messages generated by Internet Control Message Protocol (ICMP)

What IP does NOT provide End-to-end data reliability & flow control (done by TCP or application layer protocols) Sequencing of packets (like TCP) Error detection in payload (TCP, UDP or other transport layers) Error reporting (ICMP) Setting up route tables (RIP, OSPF, BGP etc) Connection setup (it is connectionless) Address/Name resolution (ARP, RARP, DNS) Configuration (BOOTP, DHCP) Multicast (IGMP, MBONE)

Internet Protocol (IP) Two versions IPv4 IPv6 IPv4 dominates today’s Internet IPv6 is used sporadically 6Bone, Internet 2

IPv4 Header 15 31 Ver HLen TOS Length Ident Flags Offset TTL Protocol 15 31 Ver HLen TOS Length Ident Flags Offset TTL Protocol Checksum SrcAddr DestAddr Options Pad

IPv4 Header Fields (1) Ver: version of protocol First thing to be determined IPv4  4, IPv6  6 Hlen: header length (in 32-bit words) Usually has a value of 5 When options are present, the value is > 5 TOS: type of service Packet precedence (3 bits) Delay/throughput/reliability specification Rarely used

IPv4 Header Fields (2) Length: length of the datagram in bytes Maximum datagram size of 65,535 bytes Ident: identifies fragments of the datagram (Ethernet 1500 Bytes max., FDDI: 4900 Bytes Max., etc.) Flag: indicates whether more fragments follow Offset: number of bytes payload is from start of original user data

Fragmentation Example 20-byte optionless IP headers Id = x 1 492 data bytes Id = x Id = x 1 492 1400 data bytes 492 data bytes Id = x 984 416 data bytes

IPv4 Header Fields (3) TTL: time to live gives the maximum number of hops for the datagram Protocol: protocol used above IP in the datagram TCP  6, UDP  17, Checksum: covers IP header

IPv4 Header Fields (4) SrcAddr: 32-bit source address DestAddr: 32-bit destination address Options: variable list of options Security: government-style markings Loose source routing: combination of source and table routing Strict source routing: specified by source Record route: where the datagram has been Options rarely used

IPv6 Initial motivation: 32-bit address space completely allocated by 2008. Additional motivation: header format helps speed processing/forwarding header changes to facilitate QoS new “anycast” address: route to “best” of several replicated servers IPv6 datagram format: fixed-length 40 byte header no fragmentation allowed (done only by source host)

IPv6: Differences from IPv4 Flow label Intended to support quality of service (QoS) 128-bit network addresses No header checksum – reduce processing time Fragmentation only by source host Extension headers Handles options (but outside the header, indicated by “Next Header” field

IPv6 Headers 15 31 Ver Pri Flow Label Payload Length Next Header 15 31 Ver Pri Flow Label Payload Length Next Header Hop Limit Source Address Destination Address

IPv6 Header Fields (1) Ver: version of protocol Pri: priority of datagram 0 = none, 1 = background traffic, 2 = unattended data transfer 4 = attended bulk transfer, 6 = interactive traffic, 7 = control traffic Flow Label Identifies an end-to-end flow IP “label switching” Experimental

IPv6 Header Fields (2) Payload Length: total length of the datagram less that of the basic IP header Next Header Identifies the protocol header that follows the basic IP header TCP => 6, UDP => 17, ICMP => 58, IP = 4, none => 59 Hop Limit: time to live

IPv6 Header Fields (3) Source/Destination Address 128-bit address space Embed world-unique link address in the lower 64 bits Address “colon” format with hexadecimal FEDC:BA98:7654:3210:FEDC:BA98:7654:3210

Addressing Modes in IPv6 Unicast Send a datagram to a single host Multicast Send copies a datagram to a group of hosts Anycast Send a datagram to the nearest in a group of hosts

Migration from IPv4 to IPv6 Interoperability with IPv4 is necessary for gradual deployment. Two mechanisms: dual stack operation: IPv6 nodes support both address types tunneling: tunnel IPv6 packets through IPv4 clouds Unfortunately there is little motivation for any one organization to move to IPv6. the challenge is the existing hosts (using IPv4 addresses) little benefit unless one can consistently use IPv6 can no longer talk to IPv4 nodes stretching address space through address translation seems to work reasonably well