Review of Networking Technologies Vahid Tabatabaee Fall 2007 ENTS689L: Packet Processing and Switching Networking Technologies 1 References Title: Internetworking with TCP/IP vol. I Principles, Protocols, and Architecture Author: Douglas E. Comer Publisher: Prentice-Hall Title: Network Processors Architectures, Protocols, and Platforms Author: Panos C. Lekkas Publisher: McGraw-Hill Title Computer Networking: A Top-Down Approach Author: J.F. Kurose, K.W. Ross Publisher: Addison Wesley ENTS689L: Packet Processing and Switching Networking Technologies 2 Connection Oriented Communication Connection oriented communication forms a dedicated connection (circuit, path) between two points. E.g. Telephone systems, ATM, Frame Relay Potential Advantages: Easier to make it reliable (note it is not necessarily more reliable). Guaranteed/reserved bandwidth. Identify flows by connection identifier rather than source/destination address (tag switching) ENTS689L: Packet Processing and Switching Networking Technologies 3 Connection-less Communication Data is segmented into packets. Every packet has identification information that enables network hardware to send it to the specified destination. E.g. Ethernet, Internet Protocol Potential Advantages: More efficient use of resources ENTS689L: Packet Processing and Switching Networking Technologies 4 Network Hierarchy (3 layer) Enterprise Network Typical networks in companies, universities. Based on Ethernet, Fast-Ethernet, WiFi. Contains one or more LAN connecting PC, printers, servers They can also have faster connections based on Gigabit Ethernet to connect to server and storage subsystems. Gateway and customer access routers provide connection to the rest of the world Access Network (provider network) Aggregate customer traffic. Send the aggregated traffic through a larger pipe into the WAN. We have three kind of routers here: Provider Edge Router (infamous last mile connection) Provider core router The typical speed range OC-3 and OC-48 WAN Interconnect provide or career networks The typical speed ranges between OC-12 and OC-192 ENTS689L: Packet Processing and Switching Networking Technologies 5 Conceptual Hierarchy of Networks WAN Access Network Provider Edge Router Enterprise Network ENTS689L: Packet Processing and Switching Networking Technologies Customer Edge Router 6 LAN/MAN/WAN Source: http://www.crema.unimi.it/didattica/Labsistemi/matagg/Tutorial%20Networking.htm ENTS689L: Packet Processing and Switching Networking Technologies 7 MAN MAN are large network spanning a campus or city. MAN (WAN) is generally less than (over) 30-50 Km. WAN spans central office facilities, while a MAN starts and ends in a central office ENTS689L: Packet Processing and Switching Networking Technologies Source: http://www.dbguide.net/know/know103001.jsp?mode=view&pg=1&idx=1038 8 4 layer Network Hierarchy WAN Core 3 Core Router Requirements: OC-192 wire-speed IP routing and MPLS VPN Traffic Engineering ENTS689L: Packet Processing and Switching Networking Technologies Edge LAN(s) Access 2 MAN CPE 1 Edge Router Requirements: Aggregate multiple access network interfaces Access network uplinks can be GigE OC-12 Reliability Redundant component Hot pluggable line cards Multiservice Providing Platform (MSSP) TDM OC-3 to OC-192 GigE, 10GigE LAN, 10GigE WAN SAN (Fibre Channel, …) IP services 9 Sprint Logical Topology 18 IP routers with 36 bidirectional logical links Source: A. Nucci, A. Sridharan, N. Taft, “The Problem of Synthetically Generating IP Traffic Matrices: Initial Recommendations”, ACM Computer Communication Review, vol. 35, no. 3, pp. 19-32 ,July 2005. ENTS689L: Packet Processing and Switching Networking Technologies 10 Sprint WDM Technology 36 OXC with 55 WDM fibers.. Wa = 40 OC-192 channels, Wb = 40 OC-48 channels, Wc = 40 OC-12 channels Source: A. Nucci, A. Sridharan, N. Taft, “The Problem of Synthetically Generating IP Traffic Matrices: Initial Recommendations”, ACM Computer Communication Review, vol. 35, no. 3, pp. 19-32 ,July 2005. ENTS689L: Packet Processing and Switching Networking Technologies 11 Ethernet Technology Ethernet is the most popular LAN technology: Shared media Carrier Sense Multiple Access/ Collision Detection (CSMA/CD) There are different variants of the Ethernet technology: Coaxial Thin wire Twisted Pair (10Base-T) ENTS689L: Packet Processing and Switching Networking Technologies 12 10Base-T Data rate : 10Mbps Broadcast, bus technology Best effort delivery: Hardware provide no information to the sender that the packet is delivered Source: http://www.webclasses.net/Courses/Intro/6.1/demo/units/unit02/sec04b.html Max. Segment length: 100m Repeaters relay electrical from one cable to another. At most two bridges between any two machines. Bridges learn addresses and replicate the signal if needed. They isolate Ethernet segments from each other. ENTS689L: Packet Processing and Switching Networking Technologies 13 Collision Detection and Recovery It is possible that two transmitters send data simultaneously and collision happens. Each transceiver monitors the cable to see if there is a coliision. When it detects collision it aborts transmission and remain idle before trying again. They use a binary exponential back-off policy. ENTS689L: Packet Processing and Switching Networking Technologies 14 Ethernet Hardware Address Ethernet defines a 48-bit addressing scheme. Each hardware card has a unique address assigned to it. Ethernet addresses are sometimes called hardware or physical addresses. Interface card receives all packets, but only send to the host that are addressed to it. Three types of address: Physical address of one network interface. The network broadcast address (all 1s) Multicast address: some interfaces can be programmed to recognize multicast addresses. ENTS689L: Packet Processing and Switching Networking Technologies 15 Ethernet Frame Format Ethernet frame size are between 64 and 1518 bytes (including header, data, and CRC). There is also 12 byte gap between Ethernet frames. Preamble 7 octets SFD 1 octet Destination Address 6 octets Source Address 6 octets Frame Type 2 octets Frame Data 64-1500 octets CRC 4 octets 10101011 Alternating 0 and 1 ENTS689L: Packet Processing and Switching Networking Technologies 16 1 GigE and 10 GigE It preserves compatibility with legacy software applications developed for running on 10BaseT. The technology has been proposed both for LAN and MAN/WAN. The jumbo frames can be up to 9000 bytes data. 10 GigE is not based on the CSMA/CD technology anymore. ENTS689L: Packet Processing and Switching Networking Technologies 17 Internet Architecture Networks are connected by routers Routers need to know about the topology of the internet beyond the networks to which they connect. Routers use the destination network, not the destination host, when routing a packet. Workstation Host Net 1 Net 2 Router 1 ENTS689L: Packet Processing and Switching Networking Technologies Net 3 Router 2 18 Questions What is the exact form of Internet Addresses? How Internet addresses are mapped to the Hardware addresses such as Ethernet addresses? ENTS689L: Packet Processing and Switching Networking Technologies 19 Host Universal Identifiers Host Identifiers are: Names: specify what an object is. Addresses: Where it is. Routes: How to get there. ENTS689L: Packet Processing and Switching Networking Technologies 20 Classes of IP Addresses Each host has a unique 32 bit internet address. Each address is a pair of (netid, hostid). 8 Class A 0 Class B 1 0 Class C 1 1 0 Class D 1 1 1 0 Class E 1 1 1 1 24 16 31 0.0.0.0 --127.255.255.255 hostid netid hostid netid netid 0 128.0.0.0 --191.255.255.255 hostid 192.0.0.0 --223.255.255.255 multicast address 224.0.0.0 --239.255.255.255 reserved for future use 240.0.0.0 --255.255.255.255 127.0.0.1 is the loopback address in IP. (127.0.0.0 to 127.255.255.255) 0.0.0.0 – 0.255.255.255 (zero addresses should not be used). 255.255.255.255 broadcast to all other nodes on the LAN In general zeros mean this and ones mean all. ENTS689L: Packet Processing and Switching Networking Technologies 21 Classless Inter-Domain Routing It replaces the older system based on classes Most sites were too big for class C and received class B number. Depletion of class B addresses (~16,000 total) It looks like a normal IP address but it ends with a slash and a number following it. It facilitates routing by allowing blocks of addresses to be grouped together into single routing table entries. Source: http://en.wikipedia.org/wiki/Classless_Inter-Domain_Routing ENTS689L: Packet Processing and Switching Networking Technologies 22 An example for CIDR Different CIDR prefixes are used in different locations for routing Outside MCI network Inside MCI network Inside ARS network ENTS689L: Packet Processing and Switching Networking Technologies 23 Weaknesses in Internet Addressing If a computer moves from one network to another, its IP address must change. This is the main source of challenge for mobile IP. The path used for hosts with multiple IP addresses (multi-homed hosts) depends on the address used. If host B connection to network 1 fails, packets from host A that uses I3 address can not reach host B. Network 1 I1 I2 I3 Router HOST A HOST B I5 I4 Network 2 ENTS689L: Packet Processing and Switching Networking Technologies 24 Address Aggregation ENTS689L: Packet Processing and Switching Networking Technologies Source: “Computer Networking: A Top-Down Approach” by J.F. Kurose, K.W. Ross 25 More specific routes ENTS689L: Packet Processing and Switching Networking Technologies Source: “Computer Networking: A Top-Down Approach” by J.F. Kurose, K.W. Ross 26 Mapping Internet Address to Physical Address Consider two machines A and B on the same network. IA, IB are internet addresses and PA, PB are physical addresses of A and B respectively. A wants to send a packet to B, but it only has IB address of B. Address Resolution Protocol (ARP) resolves this problem for networks with static address and broadcast capability. ENTS689L: Packet Processing and Switching Networking Technologies 27 ARP Host A has an ARP cache of recently acquired IP-to-physical address bindings. If IB is not in the cache, then A broadcasts an ARP request containing IB. Host B responds with an ARP reply to A that contains (IB, PB). It also adds (IA, PA) to its own cache. Sender A queues all packets destined to B until it receives ARP reply. Expiring timer for the binding entries in the cache. Resending the request (at least once) if did not get a reply. In most technologies a single type value is used for ARP frames. In Ethernet type field of (0806)16 is for ARP messages. ENTS689L: Packet Processing and Switching Networking Technologies 28 RARP A machine uses Reverse Address Resolution Protocol (RARP) to get its IP address from the server at the startup. The server has a list of IP addresses of the machines. Machine uses its physical address to communicate on the network. ENTS689L: Packet Processing and Switching Networking Technologies 29 IP Packet (Datagram) delivery The maximum size of an IP datagram is 216 octets. It is more efficient to carry each IP packet in a network frame (encapsulation). Each technology has a different maximum frame size Ethernet 1500 octets FDDI 4470 octets IP chooses a appropriate initial datagram size. Fragmentation is the process of dividing larger packets into smaller ones to adhere to the network Maximum Transfer Unit (MTU). Destination uses the Identification, Flags, and Fragment offset to reassembly the packet. ENTS689L: Packet Processing and Switching Networking Technologies 30 Time To Live field Time To Live field specifies how long a packet is allowed to be in the Internet. The source sets the maximum time that the datagram should survive. Each router decrement this field by one when it process the packet. To take into account buffering delay, each router records arrival time and decrement the field by seconds that packet stays in the router. When TTL reaches zero, the router discards the packet. ENTS689L: Packet Processing and Switching Networking Technologies 31 IP routing Direct Delivery: Transmission of packets from one machine across a SINGLE physical network to another. Indirect Delivery: Destination is not connected directly to the network of the sender, hence sender should pass the packet to a router for delivery. ENTS689L: Packet Processing and Switching Networking Technologies 32 Direct Delivery Does not involve the router. Sender encapsulates the datagram in a single physical frame. Binds the destination IP address to a physical hardware address using ARP (if needed). How does the sender know if the destination is directly connected to the same network? ENTS689L: Packet Processing and Switching Networking Technologies 33 Indirect Delivery Host (sender): Encapsulates the datagram in a physical network frame and send it to a router attached to the network. Router: Extracts the encapsulated datagram, Decides the next router to send the datagram to. Encapsulates the Datagram for transmission over the next network. Question: How a router and host decide next router to send the datagram to? ENTS689L: Packet Processing and Switching Networking Technologies 34 IP Routing Table Every host and router has a routing table Routing table has information about destination and how to reach them. We can not have a separate entry for every possible destination. First refinement: We can have a single entry for all hosts connected to the same network and only check the netid part of the internet address. Second refinement: We only need to keep the information for the next hop not the entire path for each destination. All next hop routers listed in router M routing table must lie on networks to which M connects directly. ENTS689L: Packet Processing and Switching Networking Technologies 35 Routing Table Longest Prefix Match: Recall that in CIDR more than one table entry may match the destination address. The one with largest prefix number is used for routing. Routing Table Entries: Both entries match the destination 192.168.0.0/16 address 192.168.20.19, but the second 192.168.20.16/28 one with larger prefix number is used. Default Routes: A default route is used by a router or a server when no other known route works for a packet destination address. The default route in CIDR is 0.0.0.0/0 Hosts and routers in an organization generally point the default route towards the router that has connection to a network service provider Source: Wikipedia ENTS689L: Packet Processing and Switching Networking Technologies 36 Example: Route Selection in Cisco Routers This example is based on the “Route Selection in Cisco Routers, Document ID: 8651 available at: http://www.cisco.com/en/US/tech/tk365/tec hnologies_tech_note09186a0080094823. shtml ENTS689L: Packet Processing and Switching Networking Technologies 37 Routing Table There are three process involved in building and maintaining the routing table: Various Routing Process, which run a routing protocol such as: Enhanced Interior Gateway Routing Protocol (EIGRP) Border Gateway Protocol (BGP) Intermediate System-to-Intermediate System (IS-IS) Open Shortest Path First (OSPF) The routing table, which accepts information from routing process and replies to requests from the forwarding process. The forwarding process, which requests information from the routing table for packet forwarding. ENTS689L: Packet Processing and Switching Networking Technologies 38 Building the Routing Table Main considerations in building the routing table: Administrative Distance: This indicates how much we trust source of the route. Metric: This is a measure used by the routing protocol to calculate the best path to a given destination. Prefix length ENTS689L: Packet Processing and Switching Networking Technologies 39 Routing Table Entry Update Assume there are four routing process running: EIGRP, OSPF, RIP, IGRP All 4 process learned various routes to 192.168.24.0/24 and each has chosen its best path to that network using its internal metrics and process. Each routing process attempts to install their route in the routing table. The one with lowest administrative distance (EIGRP here) can install its route in the routing table Other routes may be used as backup routes ENTS689L: Packet Processing and Switching Networking Technologies Default Administrative Distances Connected 0 Static 1 eBGP 20 EIGRP (internal) 90 IGRP 100 OSPF 110 IS-IS 115 RIP 120 EIGRP (external) 170 iBGP 200 EIGRP summary route 5 40 Prefix Lengths Assume the three routing process have received these routes: EIGRP (internal): 192.168.32.0 / 26 RIP: 192.168.32.0 / 24 OSPF: 192.168.32.0 / 19 ALL these routes will be installed in the routing table, since they have different prefix length. Routing Table: .... D 192.168.32.0/26 via 10.1.1.1 R 192.168.32.0/24 via 10.1.1.2 O 192.168.32.0/19 via 10.1.1.3 .... ENTS689L: Packet Processing and Switching Networking Technologies A packet destined for 192.168.32.1 is forwarded to 10.1.1.1, which has the longest prefix match (26 bits verses 24 or 19 bits). A packet destined for 192.168.32.100 is forwarded to 10.1.1.2, because it does NOT fall within 192.168.32.0/26 (192.168.32.0— 192.168.32.63). But it falls within the 192.168.32.0/24 destination (192.168.32.0-192.168.32.255) 41 Forwarding Decision Process ENTS689L: Packet Processing and Switching Networking Technologies 42 Dynamic Host Configuration Protocol (DHCP) The router IP addresses are typically configured manually, often remotely with a network management tool. Host addresses is typically configured using the DHCP protocol. DHCP can give a host the same IP address each time it connects to the network or assign a temporary IP address that will be different each time the host connects to the network DHCP also provides additional information such as subnet mask, address of the first hop router (default gateway) and address of the local DNS server. DHCP is also used commonly in residential access networks and in wireless LANs. ENTS689L: Packet Processing and Switching Networking Technologies 43 DHCP architecture DHCP is a client-server protocol Newly arriving hosts are clients DHCP server has the information requested by the clients Subnets may have a DHCP server If there is no server in a subnet, a DHCP relay agent knows the address of a DHCP server for that network Source:http://www.windowsitpro.com/Files/5181/Figure_02.gif ENTS689L: Packet Processing and Switching Networking Technologies 44 DHCP 4-step process DHCP server discovery: Newly arriving host sends DHCPDISCOVER message a UDP packet to port 67. This message is broadcasted. DHCP sservers respond with DHCPOFFER, which is again broadcasted. The message cotains the transaction ID, the proposed IP address, the network mask, lease time. The client will choose one server offer and respond to that server with a DHCPrequest message. The server responds with DHCPACK When the renewal timer expires When the rebinding timer expires Source:http://www.windowsitpro.com/Files/5181/Figure_01.gif ENTS689L: Packet Processing and Switching Networking Technologies 45 Mobility Management Home Network: The permanent home of a mobile node. Home Agent: The entity within the home network that performs the mobility management functions. Foreign Network: The network where the mobile node is currently residing. Foreign Agent: The entity in the foreign network that help the mobile node with the mobility management functions. Correspondent: The entity that wants to communicate with the mobile node ENTS689L: Packet Processing and Switching Networking Technologies 46 Mobile Network Architecture ENTS689L: Packet Processing and Switching Networking Technologies Source: “Computer Networking: A Top-Down Approach” by J.F. Kurose, K.W. Ross 47 Addressing (Naïve Approach) Foreign Network advertises to its neighbors that it has a highly specific route to the mobile node permanent IP address. When mobile node leaves one foreign network and joins another the new foreign network, the new foreign agent would advertise that it has a specific route and the old one eould withdraw its routing information. Drawback: Scalability; it completely destroys the hierarchical structure of IP addresses. ENTS689L: Packet Processing and Switching Networking Technologies 48 Addressing for mobile Foreign agent is located at the edge of the foreign network. Foreign agent creates a care-of address (COA) for the mobile node, with the network portion of the COA matching that of the foreign network. Note that there are two addresses for the mobile node: Permanent address Foreign address The foreign agent informs the home agent that the mobile node is in its network and has the given COA. ENTS689L: Packet Processing and Switching Networking Technologies 49 Indirect Routing to a Mobile Node The correspondent addresses the packet to the mobile node’s permanent address. The packet is first routed to the mobile node home agent. The home agent forwards the packet to a mobile node in two steps: The packet is first forwarded to the foreign agent using the COA From the foreign agent to the mobile node. Mobile node can address directly its packets to the correpondent. ENTS689L: Packet Processing and Switching Networking Technologies 50 Indirect Forwarding ENTS689L: Packet Processing and Switching Networking Technologies Source: “Computer Networking: A Top-Down Approach” by J.F. Kurose, K.W. Ross 51 Encapsulation and Decapsulation Home agent encapsulates packets Foreign agent decapsulates packets ENTS689L: Packet Processing and Switching Networking Technologies Source: “Computer Networking: A Top-Down Approach” by J.F. Kurose, K.W. Ross 52 Network Layer Functionality to Support Mobility Mobile Node to Foreign Agent: The mobile node registers with the foreign agent and deregisters when it leaves the network. Foreign Agent to Home Agent: Foreign agent registers the mobile node ‘s COA with the home agent. Does it need to deregister the COA when the mobile node leaves its network? Home Agent Packet Encapsulation: Encapsulation and forwarding of the original packets within a packet with the COA address Foreign Agent Decapsulation: Extraction of the correspondent’s original packet and forwarding of it to the mobile node. ENTS689L: Packet Processing and Switching Networking Technologies 53