WAN Technologies and Routing Strategies Updated: 11/09/20112 Large Spans and Wide Area Networks o The key issue that separates WAN technologies from LAN technologies is scalability n A WAN must be able to grow as needed to connect many sites n spread across large geographic distances o A technology is not classified as a WAN unless it can deliver reasonable performance for a large scale network n A WAN does not merely connect to many computers at many sites n It must provide sufficient capacity to permit all computers to communicate o Thus, a satellite bridge that connects a pair of PCs and printers is merely an extended LAN3 Traditional WAN Architecture o Pre-LAN WAN designers chose to create a special-purpose hardware device that could be placed at each site o A packet switch provides n local connections for computers at the site n as well as connections for data circuits that lead to other sites o A packet switch consists of a small computer system o Early packet switches were constructed from conventional computers n with a processor, memory, and I / O devices used to send and receive packets n the packet switches used in the highest-speed WANs require special-purpose hardware© 2009 Pearson Education Inc., Upper Saddle River, NJ. All rights reserved. 4 Traditional WAN Architecture Low-speed Connections High-speed Connections5 Modern WAN Architecture o Since the advent of LAN technology, most modern WANs separate a packet switch into two parts: n a Layer 2 switch that connects local computers n a router that connects to other sites6 Forming a WAN o A WAN can be formed by interconnecting a set of sites o The exact details of the interconnections depend on n the data rate needed n the distance spanned n and the delay o Many WANs use leased data circuits o A network designer must choose a topology n For a given set of sites, many topologies are possible7 Store and Forward Paradigm o WAN is to allow n as many computers as possible to send packets simultaneously o The fundamental paradigm used to achieve simultaneous transmission is known as store and forward o To perform store and forward processing n a packet switch buffers packets in memory o The store operation occurs when a packet arrives: n I / O hardware in the switch places a copy of the packet in memory o The forward operation occurs once a packet has arrived and is waiting in memory. The processor n examines the packet n determines its destination n and sends the packet over the I / O interface that leads to the destination8 Addressing in a WAN o WANs addresses follow a key concept that is used in the Internet: hierarchical addressing n Hierarchical addressing divides each address into two parts: (site, computer at the site) n In practice, instead of a identifying a site, each packet switch is assigned a unique number o first part of an address identifies a packet switch o second part identifies a specific computer n A computer connected to port 6 on packet switch 2 is assigned address [2, 6]9 Next-Hop Forwarding o What is the importance of hierarchical addressing? o When a packet arrives n a switch must choose an outgoing path over which to forward it o To make the choice, a packet switch n examines the destination address in the packet n and extracts the packet switch number o If the number in the destination address is identical to the packet switch's own ID the packet is intended for a computer on the local packet switch o Otherwise, the packet is intended for a computer on another switchNext-Hop Forwarding © 2009 Pearson Education Inc., Upper Saddle River, NJ. All rights reserved. 10 When packet with [3,5] arrives, the switch extracts 3 à send it to Interface (port) 4; as indicated by the Forwarding Table. n one entry per packet switch instead of one entry per destination computero Using only one part of a two-part hierarchical address to forward a packet has two practical consequences n First, the computation time required to forward a packet is reduced because the forwarding table can be organized as an array that uses indexing instead of searching n Second, the forwarding table contains one entry per packet switch instead of one entry per destination computer o The reduction in table size can be substantial, especially for a large WAN that has many computers attached to each packet switch o A two-part hierarchical addressing scheme allows packet switches to use only the first part of the destination address until the packet reaches the final switch n Once the packet reaches the final switch o the switch uses the second part of the address to choose a specific computer 11 Next-Hop ForwardingForwarding Mechanism and Routing o Performance n simplest is “minimum hop” n can be generalized as “least cost” o Cost is assigned based on the designed objective: delay, Queue built-up, TH, hop-count, etc. o Decision time and place n Time – when the routing decision is made o Packet (datagram) or virtual circuit basis (session based) o fixed or dynamically changing n Place – which node makes the decisions o distributed - made by each node (most common) o centralized o source o Networking information (next)Network Information Source and Update Timing o routing decisions usually based on knowledge of network (not always) n distributed routing o using local knowledge, info from adjacent nodes, info from all nodes on a potential route n central routing o collect info from all nodes o issue of update timing n how often updated? n fixed - never updated n adaptive - regular updates Topology Load Link Cost Congestion Etc. Local Information Adjacent Nodes Global Information Type DisseminationRouting Algorithm Classification (Different Views) o Global or local n Global: Entire network is known n Local: Partial knowledge about the network o Centralized or decentralized n One node maintain view of the network o Static or Dynamic n Frequent route change vs. fix routesRouting Mechanisms- How Forwarding Tables are Setup o Fixed Configuration o Flooding o