• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
[PPT]
 

[PPT]

on

  • 1,419 views

 

Statistics

Views

Total Views
1,419
Views on SlideShare
1,419
Embed Views
0

Actions

Likes
2
Downloads
7
Comments
0

0 Embeds 0

No embeds

Accessibility

Categories

Upload Details

Uploaded via as Microsoft PowerPoint

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment
  • The purpose of network layer address is to identify a device and also facilitate routing to that device. It is logical in that it is assigned by the network administrator and can be changed at will unlike the MAC address. We will see later that we need to map these logical addresses to MAC addresses for physical transmission of packets. Also, we make a distinction between routing and forwarding. Routing is about selecting a path to the destination while forwarding is about moving packets from input ports to appropriate output ports to as per the routing decisions. This distinction may seem subtle and confusing but will become clearer as we go on.
  • Here is an abstract view of the internals of a typical router. We can separate the functionality of a router into control plane and data plane. The control plane is responsible for exchanging routing information (control packets) with other routers and making routing decisions. Generally, there exists a route processor to process these control packets and to update the routing table. This routing table is then condensed into a form known as forwarding table for quick lookup by the data plane. The job of data plane is to switch packets from input ports to output ports, which is essentially a forwarding.engine. Data packets flow thru this faster data plane while control packets are processed by slower control plane. Another way to look at is that routing decisions are made by control plane and forwarding lookups done by data plane.
  • Previously we were talking about how to forward packets once the virtual circuit has been set up. For setting a VC, a source has to first select a path and send setup request along that path. We will see later how a source can get information about the network and perform path selection. Each router along the path choose a local VCI for the connection. To be precise, a downstream router selects a VCI to be used as output VCI by the upstream node. Basically we have to make sure that two distinct VCs do not have the same VCI when they flow thru the same port. Once the VCI is chosen, forwarding table is updated to reflect the new mapping from an incoming VCI and port no to outgoing VCI and port no. VC setup is essentially the updation of forwarding tables along a selected path. The key thing to note here is that VCI has only local significance and that’s why VCI of a packet is changed at each router along the path. If we want VCI to have global meaning, then we would need a larger VCI to identify every connection in the whole network. Moreover, we need to ensure that each VCI is globally unique. On the other hand, with local VCIs a router has to worry only about the VCs passing thru itself which will be much fewer. Also, some other router can also use the same VCI as long as two VCs do not get switched to the same port with same VCI. So with a smaller VCI also it is possible to have many connections in the network. Another thing to note here is the difference between routing and forwarding. Here the path selection is done the VC setup time, i.e., routing decision is made before any data is sent. And forwarding table along the selected path are updated as part of VC setup. After that each packet gets forwarded by each router/switch along the path as per the routing decision made at the time of VC setup.
  • Here is an example of how VCI translation is done. Forwarding table maps a packet from an input port with input VCI to an output port and output VCI. This forwarding table is set up such that no two packets belonging to different connections (with different input VCIs or from different input ports) get switched to the same output port with same output VCI. In this example, forwarding table says that a packet coming thru input port 1 with VCI 2 should be sent out on port 4 and VCI of the packet be changed to 1. Similarly packets from port 2 with VCI 1 are sent out on port 3 with VCI 2. You can see that no two output VCIs are same if the output port is also same.

[PPT] [PPT] Presentation Transcript

  • Topics
    • Interconnecting LAN segments
      • HUB (Physical Layer)
      • Bridge (Link layer)
      • Layer 2 Switch (multi-port bridge, link layer)
    • Interconnecting networks
      • Layer 3 Switch (network layer)
      • Router (network layer)
    • ATM Networks
  • Interconnecting LAN Segments
    • (Repeating) Hubs (layer 1 devices)
    • Bridges (layer 2 devices)
      • Basic Functions
      • Self learning and bridge forwarding table
      • Forwarding/filtering algorithm
      • Bridge looping problem and spanning tree algorithm
    • Ethernet Switches
      • Remark: switches are essentially multi-port bridges.
      • What we say about bridges also holds for switches!
    • Readings
      • Section 3.2
  • Interconnecting with Hubs
    • Backbone hub interconnects LAN segments
    • Extends max distance between nodes
    • But individual segment collision domains become one large collision domain
      • if a node in CS and a node EE transmit at same time: collision
    • Can’t interconnect 10BaseT & 100BaseT
      • Encoding is different: Manchester vs. 4B/5B
    Recreates each bit, boosts its energy strength, and transmits the bit to all other interfaces
  • Bridges
    • Link layer device
      • stores and forwards Ethernet frames
      • examines frame header and selectively forwards frame based on MAC destination address -- filtering
      • when frame is to be forwarded on a LAN segment, uses CSMA/CD to access the LAN segment
    • transparent
      • hosts are unaware of the presence of bridges
    • plug-and-play, self-learning
      • bridges do not need to be configured
  • Bridges: Traffic Isolation
    • Bridge installation breaks LAN into LAN segments
    • Bridges filter packets:
      • same-LAN-segment frames not usually forwarded onto other LAN segments
      • segments become separate collision domains
    bridge collision domain collision domain = hub = host LAN (IP network) LAN segment LAN segment
  • Forwarding How to determine to which LAN segment to forward frame?
  • Self Learning
    • A bridge has a bridge (forwarding) table
    • Entry in bridge forwarding table:
      • <Node LAN Address, Bridge Interface, Time Stamp>
      • stale entries in table dropped (TTL can be 60 min)
    • Bridges learn which hosts can be reached through which interfaces
      • when frame received, bridge “learns” location of sender: incoming LAN segment
      • records sender/location pair in bridge forwarding table
  • Filtering/Forwarding
    • When bridge receives a frame:
    • index bridge table using dest MAC address
    • 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
  • Bridge Example
    • Suppose C sends frame to D and D replies back with a frame to C.
    • Bridge receives frame from C
      • notes in bridge forwarding table that C is on interface 1
      • because D is not in table, bridge sends frame into interfaces 2 and 3
    • frame received by D
  • Bridge Learning: Example
    • D generates a frame for C, sends
    • bridge receives the frame
      • notes in bridge forwarding table that D is on interface 2
      • bridge knows C is on interface 1, so selectively forwards frame to interface 1
  • Interconnection without Backbone
    • Not recommended for two reasons:
      • - single point of failure at Computer Science hub
      • - all traffic between EE and SE must path over CS segment
  • Backbone Configuration Recommended !
  • Looping and Bridge Spanning Tree
    • for increased reliability, desirable to have redundant, alternative paths from source to dest
    • solution: organize bridges in a spanning tree by disabling subset of interfaces
    • with multiple paths, cycles result - bridges may multiply and forward frame forever
    Disabled Disabled
  • Bridge Spanning Tree Algorithm: Algorhyme
    • I think that I shall never see
    • A graph more lovely than a tree.
    • A tree whose crucial property
    • Is loop-free connectivity.
    • A tree that must be sure to span
    • So packets can reach every LAN.
    • First, the root must be selected.
    • By ID, it is elected.
    • Least cost paths from root are traced.
    • In the tree, these paths are placed.
    • A mesh is made by folks like me,
    • Then bridges find a spanning tree
    • -- Radia Perlman
  • Some Bridge Features
    • Isolates collision domains resulting in higher total max throughput
    • “ limitless” number of nodes and geographical coverage
      • Scalable? (broadcast, spanning tree algorithm…)
      • Heterogeneity (understands one type of LAN address only)
    • Can connect different Ethernet types
    • Transparent (“plug-and-play”): no configuration necessary
      • Dropping packets? Long latency? Frames reordered?
  • Ethernet Switches
    • Essentially a multi-interface bridge
    • layer 2 (frame) forwarding, filtering using LAN addresses
    • Switching: A-to-A’ and B-to-B’ simultaneously, no collisions
    • large number of interfaces
    • often: individual hosts, star-connected into switch
      • Ethernet, but no collisions!
  • Ethernet Switches
    • cut-through switching: frame forwarded from input to output port without awaiting for assembly of entire frame
      • slight reduction in latency
      • Cut-through vs. store and forward
    • combinations of shared/dedicated, 10/100/1000 Mbps interfaces
  • Not an atypical LAN (IP network) Dedicated Shared
  • A Few Words about VLAN
    • Virtual LAN (VLAN) – defined in IEEE 802.1q
      • Partition a physical LAN into several “logically separate” LANs
        • reduce broadcast traffic on physical LAN!
        • provide administrative isolation
      • Extend over a WAN (wide area network), e.g.,
      • via layer 2 tunnels (e.g., L2TP, MPLS) over IP-based WANs!
    • Two types: port-based or MAC address-based
      • each port optionally configured with a VLAN id
      • inbound packets tagged with this “VLAN” id
        • require change of data frames, carry “VLAN id” tags
        • tagged and untagged frames can co-exist
      • “ VLAN-aware” switches forward on ports part of same VLAN
    • More complex ! - require administrative configuration
      • static (“manual”) configuration
      • some configuration can be learned using GARP and GVRP protocols
      • more for info: google search on “VLAN tutorial”
  • Summary of LAN
    • Local Area Networks
      • Designed for short distance
      • Use shared media
      • Many technologies exist
    • Media Access Control: key problem!
      • Different environments/technologies-> different solutions!
    • Topology refers to general shape
      • Bus
      • Ring
      • Star
  • Summary (continued)
    • Address
      • Unique number assigned to station
      • Put in frame header
      • Recognized by hardware
    • Address forms
      • Unicast
      • Broadcast
      • Multicast
  • Summary (continued)
    • Type information
      • Describes data in frame
      • Set by sender
      • Examined by receiver
    • Frame format
      • Header contains address and type information
      • Payload contains data being sent
  • Summary (continued)
    • LAN technologies
      • Ethernet (bus)
      • Token Ring
      • FDDI (ring)
      • Wireless 802.11
    • Wiring and topology
      • Logical topology and Physical topology (wiring)
      • Hub allows
        • Star-shaped bus
        • Star-shaped ring
  • Summary (cont’d)
    • Interconnecting LAN Segments
      • (Repeating) Hubs
      • Bridges
        • Self learning and bridge forwarding table
        • Forwarding/filtering algorithm
        • Bridge looping problem and spanning tree algorithm
      • (Layer-2) Switches
        • store and forward switching
        • cut-through switching
  • Switching and Forwarding Network Layer
    • Switching and Forwarding
      • Generic Switch Architecture
      • Forwarding Tables:
        • Bridges/Layer 2 Switches; VLAN
        • Routers and Layer 3 Switches
    • Forwarding in Layer 3 (Network Layer)
      • Network Layer Functions
      • Network Service Models: VC vs. Datagram
        • ATM and IP Datagram Forwarding
    • Readings: Textbook: Chapter 3: Sections 3.1; 3.3-3.4
  • Hubs vs. Bridges vs. Routers
    • Hubs (aka Repeaters): Layer 1 devices
      • repeat (i.e., regenerate) physical signals
        • don’t understand MAC protocols!
        • LANs connected by hubs belong to same collision domain
    • Bridges (and Layer-2 Switches): Layer 2 devices
      • store and forward layer-2 frames based on MAC addresses
        • speak and obey MAC protocols
        • bridges segregate LANs into different collision domains
    • Routers (and Layer 3 Switches): Layer 3 devices
      • store and forward layer-3 packets based on network layer addresses (e.g., IP addresses)
        • rely on data link layer to deliver packets to (directly connected) next hop
        • network layer addresses are logical (i.e. virtual), need to map to MAC addresses for packet delivery
  • Switching and Forwarding
    • Function Division:
    • input interfaces (input ports):
    • perform forwarding
      • need to know to which output ports to send frames/packets
      • may enqueue packets and perform scheduling
    • switching Fabric:
      • move frames or packets from input ports to output ports
    • output interfaces (output ports):
      • may enqueue packets and perform scheduling
      • Perform MAC to transmit frames/packets to next hop
    Generic Switch Architecture Bridges and Routers: store-and forward devices! control plane
  • Input Port Functions
    • Decentralized switching :
    • given datagram dest., lookup output port using forwarding table in input port memory
    • goal: complete input port processing at ‘line speed’
    • queuing: if datagrams arrive faster than forwarding rate into switch fabric
    Physical layer: bit-level reception Data link layer: e.g., Ethernet
  • Output Ports
    • Buffering required when datagrams arrive from fabric faster than the transmission rate
    • Scheduling discipline chooses among queued datagrams for transmission
    Encapsulation)
  • Generic Switch Architecture
    • Input and output interfaces are connected through a switching fabric (backplane)
    • A backplane can be implemented by
      • shared memory
        • bridges or low capacity routers (e.g., PC-based routers)
      • shared bus
        • E.g., “low end” routers
      • point-to-point (switched) interconnection switching fabric
        • high performance switches (e.g., as used in high capacity routers
    input interface output interface Inter- connection Medium (Backplane) C C R I R O B
  • Three Types of Switching Fabrics
  • Switching Via Memory
    • First generation routers:
    • traditional computers with switching under direct control of CPU
    • packet copied to system’s memory
    • speed limited by memory bandwidth (2 bus crossings per datagram)
    Input Port Output Port Memory System Bus
  • Switching Via a Bus
    • datagram from input port memory
    • to output port memory via a shared bus
    • bus contention: switching speed limited by bus bandwidth
    • 1 Gbps bus, Cisco 1900: sufficient speed for access an enterprise routers (not regional or backbone)
  • Switching Via An Interconnection Network
    • overcome bus bandwidth limitations
    • Banyan networks, other interconnection nets initially developed to connect processors in multiprocessor
    • Advanced design: fragmenting datagram into fixed length cells, switch cells through the fabric.
    • Cisco 12000: switches Gbps through the interconnection network
  • Forwarding in Layer 3
    • Putting in context
    • What does layer-3 (network layer) do?
      • deliver packets “hop-by-hop” across a network
      • rely on layer-2 to deliver between neighboring hops
    • Key Network Layer Functions
      • Addressing: need a global (logical) addressing scheme
      • Routing: build “map” of network, find routes, …
      • Forwarding: actual delivery of packets!
    • Two basic network layer service models
      • datagram: “connectionless”
      • virtual circuit (VC): connection-oriented
  • What Does Network Layer Do?
    • End-to-end deliver packet from sending to receiving hosts, “hop-by-hop” thru network
      • A network-wide concern!
      • Involves every router, host in the network
    • Compare:
      • Transport layer
        • between two end hosts
      • Data link layer
        • over a physical link directly connecting two (or more) hosts
    network data link physical network data link physical network data link physical network data link physical network data link physical network data link physical network data link physical network data link physical application transport network data link physical application transport network data link physical
  • Network Layer Functions
    • Addressing
      • Globally unique address for each routable device
        • Logical address , unlike MAC address (as you’ve seen earlier)
      • Assigned by network operator
        • Need to map to MAC address (as you’ll see later)
    • Routing: building a “map” of network
      • Which path to use to forward packets from src to dest
    • Forwarding: delivery of packets hop by hop
      • From input port to appropriate output port in a router
      • Routing and forwarding depend on network service models: datagram vs. virtual circuit
  • Routing & Forwarding: Logical View of a Router A E D C B F 2 2 1 3 1 1 2 5 3 5
  • Network Service Model
    • Q: What service model for “channel” transporting packets from sender to receiver?
    • guaranteed bandwidth?
    • preservation of inter-packet timing (no jitter)?
    • loss-free delivery?
    • in-order delivery?
    • congestion feedback to sender?
    ? ? ? virtual circuit or datagram? The most important abstraction provided by network layer: service abstraction
  • Virtual Circuit vs. Datagram
    • Objective of both: move packets through routers from source to destination
    • Datagram Model:
      • Routing: determine next hop to each destination a priori
      • Forwarding: destination address in packet header , used at each hop to look up for next hop
        • routes may change during “session”
      • analogy: driving, asking directions at every corner gas station, or based on the road signs at every turn
    • Virtual Circuit Model:
      • Routing: determine a path from source to each destination
      • “ Call” Set-up: fixed path (“virtual circuit”) set up at “ call” setup time , remains fixed thru “call”
      • Data Forwarding: each packet carries “tag” or “label” ( virtual circuit id, VCI ), which determines next hop
      • routers maintain ”per-call” state
  • Virtual Circuit Switching
    • Explicit connection setup (and tear-down) phase
    • Subsequence packets follow same circuit
    • Sometimes called connection-oriented model
      • still packet switching, not circuit switching!
    • Analogy: phone call
    • Each switch maintains a VC table
    2 0 1 2 3 0 1 2 3 0 1 3 0 1 2 3 Host A Host B Switch 3 Switch 2 Switch 1 7 5 4 11
  • Datagram Switching
    • No connection setup phase
    • Each packet forwarded independently
    • Sometimes called connectionless model
    • Analogy: postal system
    • Each switch maintains a forwarding (routing) table
  • Forwarding Tables: VC vs. Datagram
    • Virtual Circuit Forwarding Table
    • a.k.a. VC (Translation) Table
    • (switch 1, port 2)
    • Datagram Forwarding Table
    • (switch 1)
  • More on Virtual Circuits
    • call setup/teardown for each call before data can flow
      • need special control protocol: “signaling”
      • every router on source-dest path maintains “state” ( VCI translation table ) for each passing call
      • VCI translation table at routers along the path of a call “weaving together” a “logical connection” for the call
    • link, router resources (bandwidth, buffers) may be reserved and allocated to each VC
      • to get “circuit-like” performance
    • “ source-to-dest path behaves much like telephone circuit” (but actually over packet network)
  • Virtual Circuit: Signaling Protocols
    • used to setup, maintain teardown VC
    • used in ATM, frame-relay, X.25
    • used in part of today’s Internet: Multi-Protocol Label Switching (MPLS) operated at “layer 2+1/2” (between data link layer and network layer) for “traffic engineering” purpose
    application transport network data link physical application transport network data link physical 1. Initiate call 2. incoming call 3. Accept call 4. Call connected 5. Data flow begins 6. Receive data
  • Virtual Circuit Setup/Teardown
    • Call Set-Up:
    • Source: select a path from source to destination
      • Use routing table (which provides a “map of network”)
    • Source: send VC setup request control (“signaling”) packet
      • Specify path for the call, and also the (initial) output VCI
      • perhaps also resources to be reserved, if supported
    • Each router along the path:
      • Determine output port and choose a (local) output VCI for the call
        • need to ensure that no two distinct VCs leaving the same output port have the same VCI!
      • Update VCI translation table (“forwarding table”)
        • add an entry, establishing an mapping between incoming VCI & port no. and outgoing VCI & port no. for the call
    • Call Tear-Down: similar, but remove entry instead
  • During data packet forwarding phase, input VCI is used to look up the table, and is “swapped” w/ output VCI (VCI translation, or “label swapping”) VCI translation table (aka “forwarding table”), built at call set-up phase 2 3 1 2 2 1 four “calls” going thru the router, each entry corresponding one call green call purple call blue call orange call 1 1
  • Virtual Circuit: Example 0 1 3 2 0 1 3 2 0 1 3 2 5 11 4 7 Router 3 Host B Router 2 Host A Router 1 Router 4 “ call” from host A to host B along path: host A  router 1  router 2  router 3  host B
    • each router along path maintains an entry for the call in its VCI translation table
    • the entries piece together a “logical connection” for the call
    • Exercise: write down the VCI translation table entry for the call at each router
  • Virtual Circuit Model: Pros and Cons
    • Full RTT for connection setup
      • before sending first data packet.
    • Setup request carries full destination address
      • each data packet contains only a small identifier
    • If a switch or a link in a connection fails
      • new connection needs to be established.
    • Provides opportunity to reserve resources.
  • ATM Networks
    • Asynchronous Transfer Mode
      • Single technology for handling voice,video, and data
    • Connection-oriented service using virtual circuits
      • In-sequence but unreliable
    • Cell switching using fixed-size cells: 53 bytes
      • Statistical multiplexing of cells of different circuits
    • Provide QoS guarantees/assurance
      • Variety of services such as CBR, VBR, ABR etc
  • Variable vs Fixed-Length Packets
    • No optimal length
      • if small: high header-to-data overhead
      • if large: low utilization for small messages
    • Fixed-Length easier to switch in hardware
      • simpler
      • enables parallelism
  • Big vs Small Packets
    • Small Improves Queue behavior
      • finer-grained pre-emption point for scheduling link
        • maximum packet = 4KB
        • link speed = 100Mbps
        • transmission time = 4096 x 8/100 = 327.68us
        • high priority packet may sit in the queue 327.68us
        • in contrast, 53 x 8/100 = 4.24us for ATM
      • near cut-through behavior
        • two 4KB packets arrive at same time
        • link idle for 327.68us while both arrive
        • at end of 327.68us, still have 8KB to transmit
        • in contrast, can transmit first cell after 4.24us
        • at end of 327.68us, just over 4KB left in queue
  • Big vs Small (cont)
    • Small improves latency (for voice)
      • voice digitally encoded at 64KBps (8-bit samples at 8KHz)
      • need full cell’s worth of samples before sending cell
      • example: 1000-byte cells implies 125ms per cell (too long)
      • smaller latency implies no need for echo cancellors
    • ATM Compromise: 48 bytes = (32+64)/2
  • ATM Cell Format
  • More on Cell Format
    • User-Network Interface (UNI)
      • host-to-switch format
      • GFC: Generic Flow Control (still being defined)
      • VCI: Virtual Circuit Identifier
      • VPI: Virtual Path Identifier
      • Type: management, congestion control, AAL5 (later, type field contains a user signaling bit to identify the end of a PDU )
      • CLPL Cell Loss Priority
      • HEC: Header Error Check (CRC-8)
    • Network-Network Interface (NNI)
      • switch-to-switch format
      • GFC becomes part of VPI field
  • Virtual Paths and VP Switch
    • Why use Virtual Paths (VPs)?
    • VCs of different VPs can have same VCIs
    • VPI/VCI translation
      • Cells are routed using VPI/VCI pairs in the header
    • VP Switch
      • Routing based on VPI only, VCI not translated
  • Segmentation and Reassembly
    • ATM Adaptation Layer (AAL)
      • Sets above ATM layer and below the layer with variable length frame
      • AAL 1 and 2 designed for applications that need guaranteed rate (e.g., voice, video)
      • AAL 3/4 designed for packet data
      • AAL 5 is an alternative standard for packet data
    ■ ■ ■ ■ ■ ■ AAL ATM AAL ATM
  • AAL 3/4
    • Convergence Sublayer Protocol Data Unit (CS-PDU) – encapsulation before segmentation
      • CPI: common part indicator (version field)
      • Btag/Etag: beginning and ending tag
      • BAsize: hint on amount of buffer space to allocate
      • Length: size of whole PDU
    CPI Btag BASize Pad 0 Etag Len 8 16 0 ─ 24 8 8 16 < 64 KB 8 User data
  • AAL 3/4 Cell Format
    • Add AAL 3/4 header and trailer to bring up to 48B
      • Type
        • BOM (10): beginning of message
        • COM (00): continuation of message
        • EOM (01): end of message
        • SSM (11): Single-segment message
      • SEQ: sequence of number
      • MID: multiplexing id or message id
      • Length: number of bytes of PDU in this cell
  • Encapsulation and Segmentation for AAL 3/4
  • AAL5
    • CS-PDU Format
      • pad so trailer always falls at end of ATM cell
      • Length: size of PDU (data only)
      • CRC-32 (detects missing or misordered cells)
    • Cell Format
      • end-of-PDU bit in Type field of ATM header
    CRC-32 < 64 KB 0 ─ 47 bytes 16 16 Reserved Pad Len 32 Data
  • Encapsulation and Segmentation for AAL5
  • Datagram Networks: the Internet model
    • no call setup at network layer
    • routers: no state about end-to-end connections
      • no network-level concept of “connection”
    • packets forwarded using destination host address
      • packets between same source-dest pair may take different paths, when intermediate routes change!
    application transport network data link physical application transport network data link physical 1. Send data 2. Receive data
  • Datagram Model
    • There is no round trip delay waiting for connection setup; a host can send data as soon as it is ready.
    • Source host has no way of knowing if the network is capable of delivering a packet or if the destination host is even up.
    • Since packets are treated independently , it is possible to route around link and node failures.
    • Since every packet must carry the full address of the destination, the overhead per packet is higher than for the connection-oriented model.
  • Network Layer Service Models: Network Architecture Internet ATM ATM ATM ATM Service Model best effort CBR VBR ABR UBR Bandwidth none constant rate guaranteed rate guaranteed minimum none Loss no yes yes no no Order no yes yes yes yes Timing no yes yes no no Congestion feedback no (inferred via loss) no congestion no congestion yes no Guarantees ?
    • Internet model being extended: MPLS, Diffserv
  • Datagram or VC: Why?
    • Internet
    • data exchange among computers
      • “ elastic” service, no strict timing req.
    • “ smart” end systems (computers)
      • can adapt, perform control, error recovery
      • simple inside network, complexity at “edge”
    • many link types
      • different characteristics
      • uniform service difficult
    • ATM
    • evolved from telephony
    • human conversation:
      • strict timing, reliability requirements
      • need for guaranteed service
    • “ dumb” end systems
      • telephones
      • complexity inside network
  • Forwarding and Switching Network Layer Summary
    • Switching and Forwarding
      • Generic Switch Architecture
      • Forwarding Tables:
        • Bridges/Layer 2 Switches; VLAN
        • Routers and Layer 3 Switches
    • Network Service (Forwarding) Models
      • Virtual Circuit vs. Datagram
      • Virtual Circuit Model: ATM example
        • VC set-up/tear-down
        • data forward operations
  • More on Router Architecture
    • Three Typical Architectures
    • Output queued
    • Input queued
    • Combined Input-Output queued
  • How to Speed Up Forwarding?
    • C – input/output link capacity
    • R I – maximum rate at which an input interface can send data into backplane
    • R O – maximum rate at which an output can read data from backplane
    • B – maximum aggregate backplane transfer rate
    • Back-plane speedup: B/C
    • Input speedup: R I /C
    • Output speedup: R O /C
    input interface output interface Inter- connection Medium (Backplane) C C R I R O B
  • Output Queued (OQ) Routers
    • Only output interfaces store packets
      • buffering when arrival rate via switch exceeds output line speed
      • queueing (delay) and loss due to output port buffer overflow!
    input interface output interface Backplane C R O
    • Advantages
      • Easy to design algorithms: only one congestion point
    • Disadvantages
      • Requires an output speedup of N, where N is the number of interfaces  not feasible
    B
  • Input Queued Routers: Pros & Cons
    • Advantages
      • Easy to built
        • Store packets at inputs if contention at outputs
      • Relatively easy to design algorithms
        • Only one congestion point, but not output…
        • need to implement backpressure
    • Disadvantages
      • Head-of-line (HOL) blocking
      • In general, hard to achieve high utilization
    input interface output interface Backplane C R I B R O C
  • Input Queued (IQ) Routers
    • Fabric slower than input ports combined -> queueing may occur at input queues
    • Head-of-the-Line (HOL) blocking: queued datagram at front of queue prevents others in queue from moving forward – achieve 59% of max throughput
    • queueing delay and loss due to input buffer overflow!
  • Combined Input-Output Queueing (CIOQ) Routers
    • Both input and output interfaces store packets
    • Advantages
      • Utilization 1 can be achieved with limited input/output speedup (<= 2)
    • Disadvantages
      • Harder to design algorithms
        • two congestion points
        • need to design flow control
      • An input/output speedup of 2, a CIOQ can emulate any work-conserving OQ scheduling algo.
    input interface output interface Backplane C R O R I B
  • Backplane
    • Point-to-point switch allows to simultaneously transfer a packet between any two disjoint pairs of input-output interfaces
    • Goal: come-up with a schedule that
      • Meet flow QoS requirements
      • Maximize router throughput
    • Challenges:
      • Address head-of-line blocking at inputs
      • Resolve input/output speedups contention
      • Avoid packet dropping at output if possible
    • Note: packets are fragmented in fix sized cells (why?) at inputs and reassembled at outputs
      • In Partridge et al, a cell is 64 bytes (cf. ATM, trade-offs?)
  • Head-of-Line Blocking Revisited
    • The cell at head of an input queue cannot be transferred, thus blocking the following cells
    Output 1 Output 2 Output 3 Input 1 Input 2 Input 3 Cannot be transferred because output buffer full Cannot be transferred because is blocked by red cell
  • Solution to Avoid Head-of-line Blocking
    • Maintain at each input N virtual queues, i.e., one per output
    • Need smart algorithms to schedule cell transfer to avoid input/output contentions, overflow output buffer, emulate output queuing mechanisms, ……
    Output 1 Output 2 Output 3 Input 1 Input 2 Input 3
  • Generic Architecture of a High Speed Router Today
    • Combined Input-Output Queued Architecture
      • Input/output speedup <= 2
    • Input interface
      • Perform packet forwarding (and classification)
    • Output interface
      • Perform packet (classification and) scheduling
    • Backplane
      • Point-to-point (switched) bus; speedup N
      • Schedule packet transfer from input to output