Discussions. What are the main service. Ch06 Link Layer. The Data Link Layer. Link Layer. Link Layer: Teminology. What are the services

Transcription

1 Mobile network Discussions The Link Layer Yanmin Zhu Department of Computer Science and Engineering Global ISP Home network Regional ISP Institutional network What are the services provided by the link layer? What defines a link? physical cable? How does addressing work in the link layer How do many nodes access a shared broadcast channel? Global ISP link The Data Link Layer Link Layer Our goals: understand principles behind data link layer services: error detection, correction sharing a broadcast channel: multiple access link layer addressing local area networks: Ethernet, VLNs instantiation, implementation of various link layer technologies 6. Introduction to the link layer 6. Error detection and correction 6.3 Multiple access links and protocols 6.4 Switched local area networks 6.5 Link virtualization: network as a link layer 6.6 Data center networking 3 4 Link Layer: Teminology What are the main service provided by the link layer? data transfer between neighboring network elements Terminology: hosts and routers are nodes communication channels that connect adjacent nodes along communication path are links wired links wireless links LNs layer- packet is a frame, encapsulates datagram 5 data-link layer has responsibility of transferring datagram from one node to adjacent node over a link 6

2 Link layer: context Datagram transferred by different link protocols over different links, e.g., First link: Ethernet Intermediate links: frame relay Last link: 80. Each link protocol provides different services e.g., may or may not provide rdt over link Link layer services framing, link access: encapsulate datagram into frame, adding header, trailer channel access if shared medium MC addresses used in frame headers to identify source, dest: different from IP address! reliable delivery between adjacent nodes we learned how to do this already (chapter 3)! seldom used on low bit-error link (fiber, some twisted pair) wireless links: high error rates 7 8 Link layer services (more) Where is link layer implemented? flow control: pacing between adjacent sending and receiving nodes error detection: errors caused by signal attenuation, noise. receiver detects presence of errors: signals sender for retransmission or drops frame error correction: receiver identifies and corrects bit error(s) without resorting to retransmission half-duplex and full-duplex with half duplex, nodes at both ends of link can transmit, but not at same time in each and every host link layer implemented in adaptor (aka network interface card NIC) or on a chip Ethernet card, 80. card; Ethernet chipset implements link, physical layer attaches into host s system buses combination of hardware, software, firmware application transport network link link physical cpu memory host bus controller (e.g., PCI) physical transmission network adapter card 9 0 daptors Communicating Link Layer datagram controller sending host frame sending side: encapsulates datagram in frame adds error checking bits, rdt, flow control, etc. Coding datagram datagram controller receiving host receiving side looks for errors, rdt, flow control, etc extracts datagram, passes to upper layer at receiving side decoding 6. Introduction to the link layer 6. Error detection and correction 6.3 Multiple access links and protocols 6.4 Switched local area networks 6.5 Link virtualization: network as a link layer 6.6 Data center networking

3 Error Detection Parity Checking EDC= Error Detection and Correction bits (redundancy) D = Data protected by error checking, may include header fields Single it Parity: Detect single bit errors otherwise Error detection not 00% reliable! protocol may miss some errors, but rarely larger EDC field yields better detection and correction 3 4 Parity Checking (cont ) Two Dimensional it Parity: Detect and correct single bit errors Cyclic Redundancy Check (CRC) more powerful error-detection coding view data bits, D, as a binary number widely used in practice (Ethernet, 80. WiFi, TM) Example 0 0 it stream: =b n- b n- b b b 0 Polynomial: (x)=b n- x n- + b n- x n- + + b x + b x + b 0 x 0x 0x x x Preliminary Cyclic Redundancy Check () Computing remainder x ( ) Qx ( ) Rx ( ) Gx ( ) XOR computations x ( ) Computing remainder Qx ( ) Rx ( ) Gx ( ) x 6 5 x 4 x Example x x 4 x x x x x x ( x) Q( x) G( x) R( x) ( x) R( x) Q( x) G( x) Example: x 6 5 x 4 x x x 4 x x x x x x 6 5 x x x 4 x x x x x 7 8 3

4 CRC overview How to compute R(x)? D( x) x r R( x) Q( x) G( x) Sender Given D(x) compute R(x) D(X) R(X) Receiver Noisy channel D R = remainder[. r ] G D (X) R (X) Using D (x) compute R*(x) R (x)?=r * (x) 9 0 CRC Example D: 00 G: 00 R(x): 0 Quiz Problem Data bits: 000 Generator: 0 Question: What is the value of R? Data plus redundancy to send 00 -bit data: 0 Even parity coding 0 3-repetition coding 000 Hamming Distance Definition: hamming distance (HD) is the number of positions at which the corresponding symbols are different How to characterize the performance of a code for error detection and correction? Code 00 Valid codeword Invalid codeword Valid codeword 3 Hamming distance = Hamming distance = 3 4 4

5 Minimum Hamming distance required Link Layer 6. Introduction to the link layer 6. Error detection and correction 6.3 Multiple access links and protocols 6.4 Switched local area networks 6.5 Link virtualization: network as a link layer 6.6 Data center networking 5 6 Multiple ccess Links and Protocols Two types of links : Type : point-to-point PPP for dial-up access point-to-point link between Ethernet switch and host Type : broadcast (shared wire or medium) old-fashioned Ethernet Upstream TV Cable 80. wireless LN Multiple access to a shared link Node Node Node 3 Node N Shared link Link capacity = R pbs shared wire (e.g., cabled Ethernet) shared RF (e.g., 80. WiFi) shared RF (satellite) humans at a cocktail party (shared air, acoustical) 7 8 Multiple ccess Protocols Model: () single shared broadcast channel () two or more simultaneous transmissions by nodes: interference collision if node receives two or more signals at the same time multiple access protocol (MC) n algorithm that determines how nodes share channel, i.e., determine when node can transmit What does an ideal MC protocol look like? ssumption: communication about channel sharing must use channel itself! no out-of-band channel for coordination

6 Ideal Multiple ccess Protocol MC Protocols: a taxonomy. when one node wants to transmit, it can send at rate R.. when M nodes want to transmit, each can send at average rate R/M 3. fully decentralized: no special node to coordinate transmissions no synchronization of clocks, slots 4. simple Three broad classes: Channel Partitioning divide channel into smaller pieces (time slots, frequency, code) allocate piece to node for exclusive use Random ccess channel not divided, allow collisions recover from collisions Taking turns nodes take turns nodes with more to send can take longer turns static dynamic 3 3 Channel Partitioning MC protocols: TDM TDM: time division multiple access access to channel in "rounds" each station gets fixed length slot (length = pkt trans time) in each round unused slots go idle example: 6-station LN,,3,4 have pkt, slots,5,6 idle 6-slot frame Channel Partitioning MC protocols: FDM FDM: frequency division multiple access channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go idle example: 6-station LN,,3,4 have pkt, frequency bands,5,6 idle FDM cable frequency bands Frequency Division Multiplexing Wavelength Division Multiplexing Wavelength division multiplexing. (a) The original bandwidths. (b) The bandwidths raised in frequency. (c) The multiplexed channel

7 What are drawbacks of static channel-partitioning protocols? How to overcome the drawbacks? asic idea Random ccess Protocols When a node accesses the link, it uses the whole capacity of the link! Node Node Node 3 Node N Shared link random access MC protocol specifies: how to detect collisions how to recover from collisions (e.g., via delayed retransmissions) Examples of random access MC protocols: LOH slotted LOH CSM, CSM/CD, CSM/C Link capacity = R pbs The key is to avoid collisions (i.e., nodes transmit at the same time) Pure (unslotted) LOH Pure loha efficiency Efficiency : long-run fraction of successful slots (many nodes, all with many frames to send) collision happens when: frame sent at t 0 collides with other frames sent in [t 0 -,t 0 +] P(success by given node) = P(node transmits). P(no other node transmits in [t 0 -,t 0 ]. Protocol operations P(no other node transmits in [t 0,t 0 +] = p. (-p) N-. (-p) N- when frame first arrives, a node transmit immediately If collision, immediately retransmits with probability p and waits for a frame time with probability -p; = p. (-p) (N-) choosing optimum p and then letting n -> infty... Feature: simple, no synchronization 4 = /(e) =

8 Slotted LOH Slotted loha efficiency node node node C E C S E C E S S suppose: N nodes with many frames to send, each transmits in slot with probability p Given a node, prob that has success in a slot = p(-p) N- Protocol operations when node obtains fresh frame, transmits in next slot if no collision: node can send new frame in next slot if collision: node retransmits frame in each subsequent slot with prob. p until success prob that any node has a success = Np(-p) N- Max efficiency = /e = Throughput Comparison What is the main reason for loha s low efficiency? Throughput versus offered traffic for LOH systems Transmissions without considering transmissions of other nodes! CSM (Carrier Sense Multiple ccess) CSM collisions spatial layout of nodes CSM: listen before transmit human analogy: don t interrupt others! Protocol operations () the adapter senses the channel if idle, it starts to transmit the frame immediately if not, it waits until it senses a clear channel and transmits the frame collisions can still occur: propagation delay means two nodes may not hear each other s transmission collision: entire packet transmission time wasted note: role of distance & propagation delay in determining collision probability

9 CSM/CD (Collision Detection) CSM/CD collision detection CSM/CD: carrier sensing, deferral as in CSM collision detection: measure signal strengths, compare transmitted, received signals colliding transmissions aborted, reducing channel wastage Protocol operations () the adapter senses the channel if idle, it starts to transmit the frame immediately if not, it waits until it senses a clear channel and transmits the frame () while transmitting, the adapter monitors for collisions if collision, abort immediately if not, transmission is successful (3)after aborting, it waits for a random amount of time (exponential backoff) Taking Turns MC protocols Except static partitioning protocols, can any MC protocol avoid packet collisions? channel partitioning MC protocols: share channel efficiently and fairly at high load inefficient at low load: delay in channel access, /N bandwidth allocated even if only active node! Random access MC protocols efficient at low load: single node can fully utilize channel high load: collision overhead taking turns protocols look for best of both worlds! 5 5 Taking Turns MC protocols Taking Turns MC protocols asic idea of Polling master node invites slave nodes to transmit in turn typically used with dumb slave devices data data poll master asic idea of Token passing: control token passed from one node to next sequentially. (nothing token message to send) T T slaves concerns: polling overhead latency single point of failure (master) 53 concerns: token overhead latency single point of failure (token) data 54 9

10 Summary of MC protocols channel partitioning, by time, frequency or code Time Division, Frequency Division random access LOH, S-LOH, CSM, CSM/CD Collision detection: easy in some technologies (wire), hard in others (wireless) CSM/CD used in Ethernet CSM/C used in 80. taking turns polling from central site, token passing luetooth, FDDI, IM Token Ring 55 Cable access network cable headend CMTS cable modem termination system ISP Internet frames, TV channels, control transmitted downstream at different frequencies splitter cable modem upstream Internet frames, TV control, transmitted upstream at different frequencies in time slots Downstream: multiple 40Mbps downstream (broadcast) channels single CMTS transmits into channels (no contention!) Upstream: multiple 30 Mbps upstream channels multiple access (contention!): all users contend for certain upstream channel time slots (others assigned) -56 Cable access network (cont ) cable headend CMTS MP frame for Interval [t, t] Downstream channel i Upstream channel j t t Minislots containing minislots request frames Residences with cable modems ssigned minislots containing cable modem upstream data frames DOCSIS: data over cable service interface spec FDM over upstream, downstream frequency channels TDM upstream: some slots assigned, some have contention downstream MP frame: assigns upstream slots request for upstream slots (and data) transmitted random access (binary backoff) in selected slots Link Layer 6. Introduction to the link layer 6. Error detection and correction 6.3 Multiple access links and protocols 6.4 Switched local area networks 6.5 Link virtualization: network as a link layer 6.6 Data center networking Switched Local area networks Router switch switch switch switch 6.4 Switched local area networks. Link-layer addressing and RP. Ethernet 3. Link-layer switches 4. Virtual local area networks (VLNs)

11 MC ddresses LN ddresses 3-bit IP address: network-layer address used to get datagram to destination IP subnet MC (or LN or physical or Ethernet) address: function: get frame from one interface to another physically-connected interface (same network) 48 bit MC address (for most LNs) burned in NIC ROM, also sometimes software settable Each adapter on LN has unique LN address 7-65-F LN (wired or wireless) -F D 58-3-D7-F-0-0 roadcast address = FF-FF-FF-FF-FF-FF = adapter 0C-C4--6F-E3-98 Permanent, globally unique 6 6 LN ddress (more) MC address allocation administered by IEEE manufacturer buys portion of MC address space (to assure uniqueness) analogy: (a) MC address: like Social Security Number (b) IP address: like postal address MC flat address portability can move LN card from one LN to another IP hierarchical address NOT portable address depends on IP subnet to which node is attached Why both MC and IP address? Why both MC and IP address? nswer: to keep the layers independent () The data link layer does not only serve IP protocol, including many others () if adapters use network-layer address, then the adapter needs to be reconfigured every time it moves (3) if not use MC add at all, then all received frames should be forwarded to network layer. Great overhead! How do MC address and IP address interact? 65 66

12 MC ddress interacting with IP F LN F D D7-F-0-0 0C-C4--6F-E3-98 Note: The sender,, gets s IP via DNS, but does not have s MC! RP: ddress Resolution Protocol RP: Mapping IP address to MC address Each IP node (host, router) on LN has RP table RP table: IP/MC address mappings for some LN nodes < IP address; MC address; TTL> TTL (Time To Live): time after which address mapping will be forgotten (typically 0 min) Question: When sending a packet to, how does gets s MC address? RP protocol: Same LN (network) ddressing: routing to another LN wants to send datagram to, and s MC address not in s RP table. broadcasts RP query packet, containing 's IP address dest MC address = FF-FF-FF-FF-FF-FF all machines on LN receive RP query receives RP packet, replies to with its ('s) MC address frame sent to s MC address (unicast) caches (saves) IP-to- MC address pair in its RP table until information becomes old (times out) soft state: information that times out (goes away) unless refreshed RP is plug-andplay : nodes create their RP tables without intervention from net administrator 69 walkthrough: send datagram from to via R focus on addressing at IP (datagram) and MC layer (frame) assume knows s IP address assume knows IP address of first hop router, R (how?) assume knows R s MC address (how?) C-E8-FF CC-49-DE-D0--7D R...0 E6-E F9-CD D-D-C F-54--0F 70 ddressing: routing to another LN creates IP datagram with IP source, destination creates link-layer frame with R's MC address as dest, frame contains -to- IP datagram MC src: C-E8-FF-55 MC dest: E6-E IP src:... IP dest:... ddressing: routing to another LN frame sent from to R frame received at R, datagram removed, passed up to IP MC src: C-E8-FF-55 MC dest: E6-E IP src:... IP src:... IP dest:... IP dest:... IP Eth Phy C-E8-FF-55 R F9-CD D-D-C7-56- IP Eth Phy C-E8-FF-55 IP Eth Phy R F9-CD D-D-C CC-49-DE-D0--7D...0 E6-E F-54--0F... CC-49-DE-D0--7D...0 E6-E F-54--0F 7 7

13 ddressing: routing to another LN R forwards datagram with IP source, destination R creates link-layer frame with 's MC address as dest, frame contains -to- IP datagram ddressing: routing to another LN R forwards datagram with IP source, destination R creates link-layer frame with 's MC address as dest, frame contains -to- IP datagram MC src: -3-F9-CD-06-9 MC dest: 49-D-D-C7-56- MC src: -3-F9-CD-06-9 MC dest: 49-D-D-C C-E8-FF-55 IP Eth Phy R F9-CD-06-9 IP src:... IP dest:... IP Eth Phy D-D-C C-E8-FF-55 IP Eth Phy R F9-CD-06-9 IP src:... IP dest:... IP Eth Phy D-D-C CC-49-DE-D0--7D...0 E6-E F-54--0F... CC-49-DE-D0--7D...0 E6-E F-54--0F ddressing: routing to another LN R forwards datagram with IP source, destination R creates link-layer frame with 's MC address as dest, frame contains -to- IP datagram C-E8-FF-55 R F9-CD-06-9 MC src: -3-F9-CD-06-9 MC dest: 49-D-D-C7-56- IP src:... IP dest:... IP Eth Phy D-D-C Switched local area networks. Link-layer addressing and RP. Ethernet 3. Link-layer switches 4. Virtual local area networks (VLNs)... CC-49-DE-D0--7D...0 E6-E F-54--0F Ethernet us Topology dominant wired LN technology: cheap $0 for NIC first widely used LN technology simpler, cheaper than token LNs and TM kept up with speed race: 0 Mbps 0 Gbps bus: coaxial cable Metcalfe s Ethernet sketch Old fashioned, ased on Coax us topology popular through mid 90s

14 Star topology today: star topology prevails active switch in center each spoke runs a (separate) Ethernet protocol (nodes do not collide with each other) Ethernet Frame Structure Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame switch Preamble: 7 bytes with pattern 0000 followed by one byte with pattern 000 used to synchronize receiver, sender clock rates star Ethernet Frame Structure (more) ddresses: 6 bytes if adapter receives frame with matching destination address, or with broadcast address (eg RP packet), it passes data in frame to network layer protocol otherwise, adapter discards frame Type: indicates higher layer protocol (mostly IP but others possible, e.g., Novell IPX, ppletalk) CRC: checked at receiver, if error is detected, frame is dropped Ethernet: Unreliable, connectionless connectionless: No handshaking between sending and receiving NICs unreliable: receiving NIC doesn t send acks or nacks to sending NIC stream of datagrams passed to network layer can have gaps (missing datagrams) gaps will be filled if app is using TCP otherwise, app will see gaps Ethernet s MC protocol: unslotted CSM/CD multiplexing 8 8 Ethernet CSM/CD algorithm Exponential ackoff. NIC receives datagram from network layer, creates frame. Carrier sensing If NIC senses channel idle, starts frame transmission If NIC senses channel busy, waits until channel idle, then transmits 3. If NIC transmits entire frame without detecting another transmission, NIC is done with frame! 4. If NIC detects another transmission while transmitting, aborts and sends jam signal fter aborting, NIC enters exponential backoff: after mth collision, NIC chooses K at random from {0,,,, m -}. NIC waits K 5 bit times, returns to Step

15 Ethernet s CSM/CD (more) Why Exponential? Jam Signal: make sure all other transmitters are aware of collision; 48 bits it time: 0. microsec for 0 Mbps Ethernet ; for K=03, wait time is about 50 msec See/interact with Java applet on WL Web site: highly recommended! demo Exponential ackoff: Goal: adapt retransmission attempts to estimated current load heavy load: random wait will be longer first collision: choose K from {0,}; delay is K 5 bit transmission times after second collision: choose K from {0,,,3} after ten collisions, choose K from {0,,,3,4,,03} When there are a small number of competitors, resolve completion in short time => wait a short time When there are a larger number of competitors, resolve completion in longer time When experiencing more collisions, be aware of more competitors CSM/CD efficiency T prop = max prop delay between nodes in LN t trans = time to transmit max-size frame efficiency 5 t prop /t trans efficiency goes to as t prop goes to 0 as t trans goes to infinity better performance than LOH: and simple, cheap, decentralized! 6.4 Switched local area networks. Link-layer addressing and RP. Ethernet 3. Link-layer switches 4. Virtual local area networks (VLNs) Hubs Switch Collision domain Collision domain Collision domain Hub Switch physical-layer ( dumb ) repeaters: bits coming in one link go out all other links at same rate all nodes connected to hub can collide with one another no frame buffering no CSM/CD at hub: host NICs detect collisions 89 link-layer device: smarter than hubs, take active role store, forward Ethernet frames transparent hosts are unaware of presence of switches plug-and-play, self-learning switches do not need to be configured 90 5

16 Switch: multiple simultaneous transmissions Switch Table hosts have dedicated, direct connection to switch switches buffer packets Ethernet protocol used on each incoming link, but no collisions; full duplex each link is its own collision domain switching: -to- and - to- simultaneously, without collisions not possible with dumb hub C switch with six interfaces (,,3,4,5,6) C Q: how does switch know that reachable via interface 4, reachable via interface 5? : each switch has a switch table, each entry: (MC address of host, interface to reach host, time stamp) looks like a routing table! Q: how are entries created, maintained in switch table? something like a routing protocol? C switch with six interfaces (,,3,4,5,6) C 9 9 Switch: self-learning switch learns which hosts can be reached through which interfaces when frame received, switch learns location of sender: incoming LN segment records sender/location pair in switch table C MC addr interface TTL Source: Dest: C Switch table (initially empty) Switch: frame filtering/forwarding When frame received:. record link associated with sending host. index switch table using MC dest address 3. if entry found for destination then { if dest on segment from which frame arrived then drop the frame else forward the frame on interface indicated } else flood forward on all but the interface on which the frame arrived Self-learning, forwarding: example Interconnecting switches frame destination unknown: flood destination location known: selective send C MC addr interface TTL Switch table (initially empty) Source: Dest: C 95 switches can be connected together S C S D E Q: sending from to G - how does S know to forward frame destined to F via S 4 and S 3? : self learning! (works exactly the same as in singleswitch case!) S 4 F G S3 H I 96 6

17 Self-learning multi-switch example Institutional network Suppose C sends frame to I, I responds to C S C S D E S 4 F G S3 H I to external network router mail server web server IP subnet Q: show switch tables and packet forwarding in S, S, S 3, S Switches vs. Routers both store-and-forward devices routers: network layer devices (examine network layer headers) switches are link layer devices routers maintain routing tables, implement routing algorithms switches maintain switch tables, implement filtering, learning algorithms Switches vs. Routers () Switch Router Pro Con Pro Con Plug-and-play roadcast No broadcast Not plug-andplay storms storms Topology restricted to spanning tree More general topologies Switched local area networks VLNs: motivation Router. Link-layer addressing and RP. Ethernet switch switch switch switch 3. Link-layer switches Group Group Group 3 4. Virtual local area networks (VLNs) 0 What are the problems with a hierarchical LN connecting different groups? Lack of traffic isolation among groups Security/privacy risk Management of user movement among groups 0 7

18 asic idea of VLN Port-based VLN Virtual Local rea Network switch(es) supporting VLN capabilities can be configured to define multiple virtual LNS over single physical LN infrastructure. Router switch ports grouped (by switch management software) so that single physical switch operates as multiple virtual switches Electrical Engineering (VLN ports -8) Computer Science (VLN ports 9-5) VLN3 Group 3 VLN Group VLN Group 03 Electrical Engineering (VLN ports -8) Computer Science (VLN ports 9-6) 04 Port-based VLN traffic isolation: frames to/from ports -8 can only reach ports -8 can also define VLN based on MC addresses of endpoints, rather than switch port dynamic membership: ports can be dynamically assigned among VLNs forwarding between VLNS: done via routing (just as with separate switches) in practice vendors sell combined switches plus routers router Electrical Engineering Computer Science (VLN ports -8) (VLN ports 9-5) 05 VLNS spanning multiple switches Electrical Engineering (VLN ports -8) Computer Science (VLN ports 9-5) trunk port: carries frames between VLNS defined over multiple physical switches frames forwarded within VLN between switches cannot be vanilla 80. frames (must carry VLN ID info) 80.q protocol adds/removed additional header fields for frames forwarded between trunk ports Ports,3,5 belong to EE VLN Ports 4,6,7,8 belong to CS VLN Q VLN frame format Link Layer preamble dest. address source address type data (payload) type CRC preamble dest. source data (payload) address address CRC -byte Tag Protocol Identifier Recomputed (value: 8-00) CRC 80. frame Tag Control Information ( bit VLN ID field, 3 bit priority field like IP TOS) 80.Q frame 6. Introduction to the link layer 6. Error detection and correction 6.3 Multiple access links and protocols 6.4 Switched local area networks 6.5 Link virtualization: network as a link layer 6.6 Data center networking

19 What is a link Overview of MPLS physical cable channel, for example, wireless channel between two wifi hosts connection provided by a swiched LN. connection provided by a telecommunication network The network layer is unaware of the actual forms, just a service (i.e., frame delivery between two IP devices) 09 Multiprotocol label switching (MPLS) MPLS is a packet-switched, virtual-circuit network in its own right Pedagogical viewpoint The network layer The link layer The Internet viewpoint link-layer technology for interconnecting two IP devices 0 MPLS Example Multiprotocol label switching (MPLS) R6 0 0 R7 initial goal: high-speed IP forwarding using fixed length label (instead of IP address) fast lookup using fixed length identifier (rather than shortest prefix matching) borrowing ideas from Virtual Circuit (VC) approach but IP datagram still keeps IP address! R5 R4 R R3 0 R 0 R8 PPP or Ethernet header MPLS header IP header remainder of link-layer frame MPLS network virtual-circuit network label Exp S TTL MPLS forwarding tables in out out label label dest interface 0 0 D 0 8 label-switched router forward packets to outgoing interface based only on label value (don t inspect IP address) MPLS forwarding table distinct from IP forwarding tables in out out label label dest interface D 0 MPLS signaling modify OSPF, IS-IS link-state flooding protocols to carry info used by MPLS routing, e.g., link bandwidth, amount of reserved link bandwidth entry MPLS router uses RSVP-TE signaling protocol to set up MPLS forwarding at downstream routers R6 R5 R4 R3 pp Transport Network 0 MPLS Link Physical R in out out label label dest interface D in out out R label label dest interface pp Transport Network MPLS Link Physical 3 R6 R5 R4 modified link state flooding RSVP-TE D 4 9

20 MPLS versus IP paths R6 R5 R4 R R3 flexibility: MPLS forwarding decisions can differ from those of IP: use destination and source addresses to route flows to same destination differently (traffic engineering) D MPLS versus IP paths R6 R5 R4 entry router (R4) can use different MPLS routes to based, e.g., on source address R R3 D IP router IP routing: path to destination determined by destination address alone 5 IP routing: path to destination determined by destination address alone MPLS routing: path to destination can be based on source and dest. address fast reroute: precompute backup routes in case of link failure IP-only router MPLS and IP router 6 Link Layer Data center networks 6. Introduction to the link layer 6. Error detection and correction 6.3 Multiple access links and protocols 6.4 Switched local area networks 6.5 Link virtualization: network as a link layer 6.6 Data center networking 0 s to 00 s of thousands of hosts, often closely coupled, in close proximity: e-business (e.g. mazon) content-servers (e.g., YouTube, kamai, pple, Microsoft) search engines, data mining (e.g., Google) Inside a 40-ft Microsoft container, Chicago data center 7 8 Typical rchitecture TOR switches ccess router Internet order router Two types of traffic ccess router Tier- switches Tier- switches What are the design objectives of a data center network? Server racks 9 0 0

21 Rich interconnection among switches increased throughput between racks (multiple routing paths possible) increased reliability via redundancy Tier- switches Tier- switches Summary principles behind data link layer services: error detection, correction sharing a broadcast channel: multiple access link layer addressing instantiation and implementation of various link layer technologies Ethernet switched LNS TOR switches Server racks let s take a breath journey down protocol stack complete (except PHY) solid understanding of networking principles, practice Lab #5 Exploring Ethernet (IEEE80.3) No submission 3 4