Lecture 4 - Switched Networks and ATM

ATM Basic Concepts

1) Negotiated Service Connection - End-to-end connections, called virtual circuits - Traffic contract 2) Virtual circuit based switching - Dedicated capacity 3) Cell Based - Small, fixed length

ATM Interfaces

An ATM network consists of a set of ATM switches interconnected by point-to-point ATM links or interfaces. ATM switches support two primary types of interfaces: UNI and NNI. The UNI connects ATM end systems (such as hosts and routers) to an ATM switch. The NNI connects two ATM switches. Depending on whether the switch is owned and located at the customer's premises or is publicly owned and operated by the telephone company, UNI and NNI can be further subdivided into public and private UNIs and NNIs. A private UNI connects an ATM endpoint and a private ATM switch. Its public counterpart connects an ATM endpoint or private switch to a public switch. A private NNI connects two ATM switches within the same private organization. A public one connects two ATM switches within the same public organization.

IP over ATM

Classic IP - Layer 3 "networks" - connect LANs - MAC (802.3) and IP addresses IP over ATM - Replace LAN segments with ATM network - ATM addresses, IP addresses Packet journey in IP-over-ATM network at Source Host (IP-over-ATM router): IP layer maps between IP and ATM dest address - IP packet into ATM AAL5 PDUs - from IP addresses to ATM addresses just like IP addresses to 802.3 MAC addresses (ARP) passes datagram to AAL5 AAL5 encapsulates data, segments cells, passes to ATM layer ATM network: moves cell along VC to destination at Destination Host (IP-over-ATM router): - AAL5 reassembles cells into original datagram - if CRC OK, packet is passed to IP

Spanning Tree Protocol (STP)

Each bridge has a unique ID (MAC addr + priority level) Select the bridge with the smallest ID as the root of the spanning tree, called "root bridge" All the ports on the root bridge are active (forwards the frames) Each bridge determines the minimum-cost path from itself to the root and nodes which of its port is on the path (root port) Link cost: the cost of traversing a single network segment (link) Path cost: the sum of the costs of the segments (links) on the path - an administrator can configure the cost of traversing a particular segment (link) - E.g. set the cost for every segment to 1, the path cost is a count of the number of bridges along the path. Root path cost: the cost of the minimum-cost path from this bridge to the root Root port: the port connecting to the minimum-cost path on this bridge Breaking ties: When multiple paths from a bridge are min-cost paths, choose the path using the neighbor bridge with the lower bridge ID. If the multiple ports connects this bridge and the neighbor bridge on the root path, choose the port with the lowest port ID as the root port. Select a single "designated bridge" and its designated port on each LAN segment Designated bridge: the bridge on that LAN segment with the minimum-cost path to the root. Only designated bridge allowed to forward frames to and from this LAN segment. - 2If two or more bridges have the same root path cost, choose the one with the lowest bridge ID Designated port: the port connecting the designated bridge to this LAN segment - If the designated bridges has two or more ports attached to this LAN, choose the port with the lowest port ID Any port that is not a root port or a designated port is blocked. Bridges exchange messages to configure the bridge (Configuration Bridge Protocol Data Unit, CBPDUs) to cut the loop and build the tree. - Source addr: port MAC addr, Dest. addr: STP multicast address - <sending bridge ID, root bridge ID, root path cost> At the beginning, each bridge considers itself to be the root, sends CBPDU identifying itself as root Upon receiving a CBPDU, check if the new path is better - if better, update its STP record, forward the message after updating the root path cost in the message - After stabilization, only the root bridge generates new CBPDUs regularly, others stops generate CBPDUs once learning it is not a root From a non-root port, receives a CBPDU indicating it is not the designated bridge for that segment, goes to blocking state - BPDU is still received in blocking state.

ATM - Negotiated Service Connection

QOS = Quality of Service QOS for the different channels of the virtual connection can be set (part of negotiating the connection in ATM) Traffic Contract: Parameters - Traffic Characteristics - Peak Cell Rate - Sustainable - Cell Rate Quality of Service - Delay - Cell Loss

ATM Physical Sublayers

Transmission Convergence Sublayer (TCS): adapts ATM layer above to PMD sublayer below - Specific to the PMD - Cell delineation - Cell rate decoupling, inserting idle (empty) cells when no data cells to send (with "unstructured" PMD sublayer) Physical Layer Medium Dependent Sublayer (PMD): depends on physical medium being used - Probably use existing standards and technology - Medium, line code, connectors Physical Medium Dependent (PMD) sublayer SONET/SDH: transmission frame structure (like a container carrying bits); - bit synchronization; - bandwidth partitions (TDM); - several speeds: OC3 = 155.52 Mbps; OC12 = 622.08 Mbps; OC48 = 2.45 Gbps, OC192 = 9.6 Gbps TI/T3:transmission frame structure (old telephone hierarchy): 1.5 Mbps/ 45 Mbps unstructured: just cells (busy/idle)

Is ATM on the Link Layer or Network Layer? (Vision vs. Reality)

Vision: provide the end-to-end transport: "ATM from desktop to desktop" ATM is network technology Reality: used to connect IP backbone routers "IP over ATM" ATM as switched link layer, connecting IP routers

Forwarding

Which port to forward a frame? - Use forwarding database/table - < MAC address, port, Time-to-Live (TTL)> How to build the forwarding table??? - A routing problem

AAL5 (AAL Protocol)

48 Bytes of Data per Cell Uses a paidload type identifier (PTI) bit in the ATM header to Indicate Last Cell Only One PDU at a Time on a Virtual Connection CRC-32: Per PDU CRC for error checking2

Transparent Bridges - Spanning Trees

A solution is to prevent loops in the topology IEEE 802.1d has an algorithm that organizes the bridges as spanning tree in a dynamic environment Note: Trees don't have loops Bridges that run 802.1d are called transparent bridges

Learning Bridge

A transparent bridge examines a packet's Destination Data Link Address and looks up the Address in its internal tables to determine which of its ports, if any, to forward the packet onto. A learning bridge can be added to a network, and it will learn the network topology without help from humans. If a station is moved, the bridge will realize it and update its tables appropriately. Originally, there were some transparent bridges that did not have learning capability, but today, the terms "transparent bridge" and "learning bridge" are used interchangeably.

Virtual Circuit Switching

Establish connection (virtual circuit) before any data is sent Permanent Virtual Circuit (PVC), manually or setup signaling initiated by the network administrator, - Long lasting connections, e.g. "permanent" coonections for two IP routers Switched Virtual Circuit (SVC), setup using signaling by one of the hosts - Dynamically set up on per-call basis Negotiate QoS (bandwidth, delay, etc) link,switch resources (bandwidth, buffers) may be allocated to VC: to get circuit-like performance Each switch on source-destination path maintains "connection state" for each passing connection - Incoming interface, incoming virtual circuit identifier (VCI), outgoing interface, outgoing VCI, reserved bandwidth, buffer, delay... Tear down Forwarding: each cell/packet carries VC identifier (not destination ID)

Switched Virtual Circuits

Established only when needed.

Ethernet Hub

Hub is just a repeater - Receive signal from one port and broadcast to all other ports Extends max distance between nodes, but collisions are propagated - Each port on a hub is in the same collision domain - Individual segment collision domains become one large collision domain Cannot interconnect different LAN technologies, e.g. 10BaseT & 100BaseT

Transparent Bridges - Frame Forwarding and Filtering

Layer 2 switches (bridges) have a MAC address table that contains a MAC address and port number. Switches follow this simple algorithm for forwarding packets: 1) When a frame is received, the switch compares the SOURCE MAC address to the MAC address table. If the SOURCE is unknown, the switch adds it to the table along with the port number the packet was received on. In this way, the switch learns the MAC address and port of every transmitting device. 2) The switch then compares the DESTINATION MAC address with the table. If there is an entry, the switch forwards the frame out the associated port. If the Destination is the same port from which the frame arrived, drop the frame (filtering). If there is no entry, the switch sends the frame out all its ports, except the port that the frame was received on (Flooding). Note that the switch does not learn the destination MAC until it receives a frame from that device.

Bridges/LAN Switches

Link layer device stores and forwards frames examines frame header and selectively forwards frame based on MAC destination address (this is learning bridge)* when frame is to be forwarded on segment, uses the corresponding MAC to access segment (e.g. CSMA/CD for Ethernet) Interconnect multiple LANs, possibly even support different IEEE 802.x types, e.g. 802.3 and 802.5, 802.11, but NOT 802.x with ATM

ATM Adaption Layer (AAL)

Only at edge of ATM network (end system) Roughly analogous to Internet transport layer Provides mapping Of applications (IP or native ATM applications) to ATM service of the same type Segments/Reassembles into 48 Payloads Hands 48 Byte Payloads To ATM Layer

Reading and Practice

Peterson & Davie, Chapter 3 Peterson & Davie, Chap 3, 4th ed -3.1 -3.5 -3.7 -3.8 -3.13 -3.26 Vol 5: 3.1,5,7,8,13, (ATM AAL5 problem on next page) Download and browse ATM UNI4.0 spec

ATM Addressing

Public networks - E.164 numbers (telephone numbers) - Up to 15 digits Private networks - 20 byte address - Format modeled after OSI NSAP (Network Service Access Point) - Mechanisms for administration exist - Hierarchical structure will facilitate virtual connection routing in large ATM networks - MAC address will be encapsulated within NSAP

Homework

Q26 from vol. 4 The IP datagram for a TCP ACK message is 40B long. It contains 20B TCP header and 20B IP header. Assume that this ACK is traversing an ATM network that uses AAL5. How many ATM cells will be needed to carry the ACK. What if AAL3/4 is used instead?

ATM Signaling

See slides, for full connection steps.

ATM Cell

Small Size (low delay, but high overhead) - 5 Byte Header - 48 Byte Payload Fixed Size (easy switch implementation, but padding overhead) Header contains virtual circuit information Payload can be voice, video or other data types

ATM Adaption Layer (AAL) and the 5 Possible AAL Protocols

The following ATM Adaptation Layer protocols (AALs) have been defined by the ITU-T.[1] It is meant that these AALs will meet a variety of needs. The classification is based on whether a timing relationship must be maintained between source and destination, whether the application requires a constant bit rate, and whether the transfer is connection oriented or connectionless. AAL Type 0 (also referred as raw cells) consists of 48 bytes of payload without any reservation for special fields. AAL Type 1 supports constant bit rate (CBR), synchronous, connection oriented traffic. Examples include T1 (DS1), E1, and x64 kbit/s emulation. AAL Type 2 supports time-dependent Variable Bit Rate (VBR-RT) of connection-oriented, synchronous traffic. Examples include Voice over ATM. AAL2 is also widely used in wireless applications due to the capability of multiplexing voice packets from different users on a single ATM connection. AAL Type 3/4 supports VBR, data traffic, connection-oriented, asynchronous traffic (e.g. X.25 data) or connectionless packet data (e.g. SMDS traffic) with an additional 4-byte header in the information payload of the cell. Examples include Frame Relay and X.25. AAL Type 5 is similar to AAL 3/4 with a simplified information header scheme. This AAL assumes that the data is sequential from the end user and uses the Payload Type Indicator (PTI) bit to indicate the last cell in a transmission. Examples of services that use AAL 5 are classic IP over ATM, Ethernet Over ATM, SMDS, and LAN Emulation (LANE). AAL 5 is a widely used ATM

5 Steps for Setting up a Call

(1) Setup message - Call reference - Called party address - Calling party address - Traffic characteristics - Quality of service Call proceeding message - Call reference - VPI/VCI (2) Internal network processing - Resource availability checking - Virtual channel or path routing - Function of the Network Node Interface (NNI) (3) Call Proceeding - Call reference Called user deciding to accept call (4) Connect message - Call reference - Indicates call acceptance Connect Acknowledge - Sent to destination host - Call reference (5) Calling party informed that call is available for user information exchange See slides, can't understand this without diagrams.

Goal of Asynchronous Transfer Mode (ATM) Network

990's standards for high-speed (155Mbps to 622 Mbps and higher) Broadband Integrated Service Digital Network (BISDN) architecture Goal: integrated, end-end transport of carry voice, video, data meeting timing/quality of service (QoS) requirements of voice, video (versus Internet best-effort model) "next generation" telephony: technical roots in telephone world packet-switching (fixed length packets, called "cells") using virtual circuits ATM network moves cells (fixed length packets) with low delay and low delay variation at high speeds Devices at ends translate (e.g., segment and reassemble) between cells and original traffic

ATM UNI Cell

A UNI interface (link) connects ATM end systems (such as hosts and routers) to an ATM switch. A UNI cell is sent over a UNI interface.

Broadcast Domain

A broadcast domain is a logical division of a computer network, in which all nodes can reach each other by broadcast at the data link layer. A broadcast domain can be within the same LAN segment or it can be bridged to other LAN segments. In terms of current popular technologies, any computer connected to the same Ethernet repeater or switch is a member of the same broadcast domain. Further, any computer connected to the same set of inter-connected switches/repeaters is a member of the same broadcast domain. Routers and other higher-layer devices form boundaries between broadcast domains. This is as compared to a collision domain, which would be all nodes on the same set of inter-connected repeaters, divided by switches and learning bridges. Collision domains are generally smaller than, and contained within, broadcast domains.

Collision Domain

A collision domain is a network segment connected by a shared medium or through repeaters where data packets may collide with one another while being sent. The collision domain applies particularly in wireless networks, but also affected early versions of Ethernet. A network collision occurs when more than one device attempts to send a packet on a network segment at the same time. Members of a collision domain may be involved in collisions with one another. Devices outside the collision domain do not have collisions with those inside.

Switch vs. Bridge

A network is formed when two or more devices connect to share data or resources. A large network may need to be subdivided for efficient frame delivery or the traffic management. Bridges or switches are used to connect these subdivided segments of networks. In a long way, the terms bridge and switch are use interchangeably. Bridge and switch both provide the same functionality but the switch does it with greater efficiency. A bridge connects smaller network segments to form a large network, and it also relays frame from one LAN to another LAN. On the other hand, the switch connects more network segments as compared to the bridges. This is a basic difference between bridge and switch.

Switch

A network switch (also called switching hub, bridging hub, officially MAC bridge[1]) is a computer networking device that connects devices together on a computer network by using packet switching to receive, process, and forward data to the destination device. In networks the switch is the device that filters and forwards packets between LAN segments. Switches operate at the data link layer (layer 2) and sometimes the network layer (layer 3) of the OSI Reference Model and therefore support any packet protocol. LANs that use switches to join segments are called switched LANs or, in the case of Ethernet networks, switched Ethernet LANs.

Virtual LAN

A virtual LAN (VLAN) is any broadcast domain that is partitioned and isolated in a computer network at the data link layer (OSI layer 2).[1][2] LAN is the abbreviation for local area network and in this context virtual refers to a physical object recreated and altered by additional logic. VLANs work by applying tags to network packets and handling these tags in networking systems - creating the appearance and functionality of network traffic that is physically on a single network but acts as if it is split between separate networks. In this way, VLANs can keep network applications separate despite being connected to the same physical network, and without requiring multiple sets of cabling and networking devices to be deployed. Group the stations in a broadcast domain, regardless of their physical location. A VLAN ID (VID) in the frame A frame is not forwarded/broadcasted from one VLAN to another VLAN Each VLAN establishes its own spanning tree Assign a port to one or multiple or all VLANs (static or dynamic)

ATM Cell Format

A) Generic Flow Control: Used for UNI only - Not NNI Currently undefined Set to 0000B Proposed future uses - Flow control - Shared media multiple access B) Payload Type Identifier (PTI): Bit 3: Used to discriminate data cells from operation, administration, maintenance cells. Bit 2: Used to indicate congestion in data cells (Bit 3 = 0) - Set by Switches - Source and Destination Behavior Defined for Available Bit Rate Flow Control VCC's Bit 1: Carried transparently end-to-end in data cells - Used by AAL5 C) Cell Loss Priority: Cells with bit set (CLP =1) should be discarded before those with bit not set (low priority) Can be set by the terminal Can be set by ATM switches for internal network control - Virtual channels/paths with low quality of service - Cells that violate traffic management contract Key to ATM Traffic Management D) Header Error Check: Header error control Detection mode: - Protects header only (all five bytes) - Discards cell when header error Correction mode (optional): Correct 1 bit errors else discard when error detected - Reduced cell loss in face of single bit errors - Reduced error detection for multiple bit errors Cell delineation for SONET, SDH, etc... Recalculated link-by-link because of VPI/VCI value changes * Entire cell is 53 bytes - 5 byte header - 48 byte payload - Size is because of compromise reached in ITU-TS Study Group XVIII in June 1989

AAL1 (Adaptive Clock Method) (AAL Protocol)

AAL1: for constant bit rate (CBR) services, e.g. circuit emulation Bit stream rate is independent of ATM network and (theoretically) can be any value Cell delay variation is critical to buffer sizing and bit clock jitter

AAL2 (AAL Protocol) (AAL Protocol)

AAL2 AAL2: variable bit rate (VBR) services, e.g. MPEG video Emulation small payload to reduce packetization delay - One cell can carry data from multiple users

AAL3/4 (AAL Protocol)

AAL3/4 for data (e.g. IP datagrams) 44 Bytes of Data per Cell Type: the first, last, middle or single cell SEQ: sequence # of the cell (to detect the cell loss) Message Identifier (MID) multiplex several PDUs onto a single virtual connection Len: length = # of bytes of PDU in the cell CRC-10: Checking per Cell

ATM Layer

Adds/Removes Header To 48 Byte Payload - Header Contains Connection Identifier, multiplexes 53 Byte cells into virtual connections, ATM's "Network" layer - Transport cells across ATM network (analogous to IP network layer, but very different strategy and services than IP network layer) - Signaling, cell switching, routing

ATM Virtual Circuits (ATM VCs)

Advantages of ATM VC approach: - QoS performance guarantee for connection mapped to VC (bandwidth, delay, delay jitter) Drawbacks of ATM VC approach: - Inefficient support of datagram traffic - one PVC between each source/dest (pair), does not scale - SVC introduces call setup latency, processing overhead for short lived connections

Ethernet Hubs vs. Ethernet Switches

An Ethernet switch is a packet switch for Ethernet frames - Buffering of frames prevents collisions - Each port is isolated and builds its own collision domain - Break subnet into LAN segments - Host can directly connect to switch, no collision, full duplex An Ethernet Hub does not perform buffering: - Collisions occur if two frames arrive at the same time.

ATM NNI Cell

An NNI interface (link) connects two ATM switches. An NNI cell is sent over an NNI interface. Supports 212 Virtual Paths Supports virtual connection routing - Distribution of topology information - Distribution of resource availability information Public version being standardized by ITU TS Private version specified by ATM Forum Technical Working Group

Switching Techniques

Build a large network by interconnecting a number of switches Easily add new hosts Switching Techniques: Datagram or connectionless (Ethernet) - Unique address - No need to setup connection Virtual circuit or connection-oriented (ATM) - Set up connection and maintain connection state Source routing - Source specify the whole or partial route to the destination

Virtual Paths and Virtual Channels

Bundles of Virtual Channels are switched via Virtual Paths Better scalability (i.e. more capable of growing to large numbers of circuits)

Call Control Signaling (ATM)

Call control protocol (Q.2931) is used to establish, maintain, and clear virtual channel connections between a user and network

Transparent Bridges - Danger of Loops

Consider the two LANs that are connected by two bridges. Assume host n is transmitting a frame F with unknown destination. What is happening? Bridges A and B flood the frame to LAN 2. Bridge B sees F on LAN 2 (with unknown destination), and copies the frame back to LAN 1 Bridge A does the same. The copying continues Solution? Transparent bridges use spanning trees to prevent loops in the topology.

ATM Service Categories

Constant Bit Rate (CBR) - Continuous flow of data with tight bounds on delay and delay variation Real-Time Variable Bit Rate (rt-VBR) - Variable bandwidth with tight bounds on delay and delay variation Non-Real-Time Variable Bit Rate (nrt-VBR) - Variable bandwidth with tight bound on cell loss Available Bit Rate (ABR) - guarantee minimum - Flow control on source with tight bound on cell loss Unspecified Bit Rate (UBR) - No guarantees (i.e., best effort delivery)

Queuing Delay Advantage of Small Cells (ATM)

Delay and delay variation are small for small messages e.g., a digitized voice sample But high header overhead

Ethernet Switching vs. Virtual Circuit Switching

Ethernet Switching: No connection setup (connection less) Packet carries dest. addr. Switching based on globally unique MAC address a host does not know whether the network is capable of delivering the packet when it sends the packet Each packet is forwarded independently and may be out of order A switch and link failure might not have any serious effect if it is possible to find an alternate route --- Virtual Circuit Switching: Establish connection state before sending any data (connection oriented) - Setup latency, processing overhead, scalability (capability to grow to a large network) Packet/cell carries VCI Switching based on incoming port + VCI (unique per port) - VCI changed at the output port Negotiate the QoS parameters and allocate resources (buffer, bandwidth) to VC - If not enough resource, reject the connection request - QoS performance guranteed for connection (bandwidth, delay, delay jitter) Each cell is routed along the established connection in order If a switch or a link fail, tear down the old connection and establish a new connection ***Many ATM ideas adopted in IP networks called MPLS

Ethernet Switching and Full-Duplex

Ethernet switching gave rise to another advancement, full-duplex Ethernet. Full-duplex is a data communications term that refers to the ability to send and receive data at the same time. Legacy Ethernet is half-duplex, meaning information can move in only one direction at a time. In a totally switched network, nodes only communicate with the switch and never directly with each other. Switched networks also employ either twisted pair or fiber optic cabling, both of which use separate conductors for sending and receiving data. In this type of environment, Ethernet stations can forgo the collision detection process and transmit at will, since they are the only potential devices that can access the medium. This allows end stations to transmit to the switch at the same time that the switch transmits to them, achieving a collision-free environment.

Transparent Bridges - Self Learning (Switch Forwarding Table)

Forwarding tables entries are set automatically with a simple heuristic: - The source address field of a frame that arrives on a port tells which host is reachable from this port (the host that sent the packet must be reachable from this port). When a frame received, switch "learns" location of sender records sender/location pair in forwarding table with TTL = MAX_TTL - TTL reset to MAX_TTL every time a frame with the same source addr is received to refresh the existing table entry - Entry removed when TTL counts down to 0 Note that the switch does not learn the destination MAC until it receives a frame from that device.

Transparent Bridges

The inability to allow more than one device to transmit simultaneously presents a major challenge when attempting to connect dozens or hundreds of users together through Ethernet. Transparent bridging is the augmentation of Ethernet allowing partial segmentation of the network into two or more collision domains. The IEEE-defined transparent bridging is an industry standard in 802.1D. Transparent bridges improve network performance by allowing devices in the same segmented collision domain to communicate without that traffic unnecessarily being forwarded to the other collision domain. Transparent bridges are the predominant bridge type for Ethernet, and it is important to understand Ethernet switches essentially act as multiport transparent bridges. Figure 1-10 shows a transparent bridge supporting Ethernet segments or collision domains. If Host1 and Host2 are talking to each other, their conversation will use bandwidth only on their side of the bridge. This allows Host4 and Host5 to also hold a conversation. If all devices were in the same collision domain, only one conversation would be possible. However, if Host1 wants to talk to Host4, as shown in Figure 1-11, the bandwidth will be utilized on both sides of the bridge, allowing only the one conversation. Three parts to transparent bridges: (1) Learning of Addresses (2) Forwarding of Frames (3) Spanning Tree Algorithm