Routers

Purpose of Routers

The purpose of a router is to connect nodes across an internetwork regardless of the Physical Layer and Data Link Layer protocol used. Routers are hardware and topology independent. Routers are not aware of the type of medium or frame used (Ethernet, Token Ring, FDDI, X.25, etc...). Routers are aware of the Network Layer protocol used: Novell's IPX, Unix's IP, XNS, Apples DDP, etc..

Router OSI Operating Layer

Routers operate on the OSI Model's Network Layer. The internetwork must use the same Network Layer protocol. Routers allow the transportation of the Network Layer PDU through the internetwork even though the Physical and Data Link Frame size and addressing scheme may change.

Router Segment to Segment Characteristics

Routers that only know Novell IPX (Internetwork Packet Exchange) will not forward Unix's IP (Internetwork Packet) PDUs and vice versa. Routers only see the Network Layer protocol that they have been configured for. This means that a network can have multiple protocols running on it: SPX/IPX, TCP/IP, Appletalk, XNS, etc..

In the following network, Router #3 is a Novell SPX/IPX router, it only sees the Network Layer protocol IPX. This means that any TCP/IP PDUs will not pass through, the router does not recognize the PDUs and doesn't know what to do with them.

Routers #1 and #2 are TCP/IP routers, they recognize only IP protocols. This keeps SPX/IPX traffic off of "Segment 300". This is in quotations because TCP/IP has a different network numbering scheme than IPX.

Important Point: Routers allow network traffic to be isolated or segmented based on the Network Layer Protocol and by the network address. This provides a functional segmentation of the network.

Routers that only can see 1 protocol are called Protocol Dependent Routers. Routers that can see many different protocols (2 or more) are called Multiprotocol Routers.

Router Addressing

Routers combine the Network Number and the Node Address to make Source and Destination addresses in routing Network Layer PDUs across an network. Routers have to know the name of the segment that they are on and the segment name or number where the PDU is going to. They also have to know the Node Address: MAC Address for Novell and the IP address for TCP/IP.

For Novell's SPX/IPX (Sequential Packet eXchange/Internetwork Packet eXchange), the Network Layer PDUs address is composed of the Network Address (32 bit number) and the Host address (48 bit - MAC address).

Routing Protocols

Routing Protocols are a "sub-protocol" of the Network Layer Protocol that deal specifically with routing of packets from the source to the destination across an internetwork. Examples of Routing Protocols are: RIP, IGRP and OSPF.

RIP - Routing Information Protocol

RIP was one of the first routing protocols to gain widespread acceptance. It is described in RFC1058 which is an Internet standard. RFC stands for request for comment and the RFC1058 is the 1,058 RFC standard published. Commercial NOS such as Novell, Apple, Banyan Vines and 3Com, use RIP as the base routing algorithm for their respective protocol suites.

RIP is a distance vector algorithm. Routers maintain a detailed view of locally attached network segments and a partial view of the remainder of the routing table. The routers contain information on the number of hop counts to each segment. A hop is considered to be one transverse through a router. Pass through a router and the Hop count increases by 1.

The routers are updated every 30 seconds, each router sending out a RIP broadcast. This advertisement process is what enables RIP routing to be dynamic. Dynamic routers can change routing tables on the fly as the network configuration changes. By using the Hop Count information from their routing tables, routers can select the shortest path - the least number of hops to the destination.

Apple uses RTMP (routing table maintenance protocol) which adds a route status indicator: good, bad or suspect depending on the age of the route information.

Novell adds ticks to the RIP algorithm, Ticks are dynamically assigned values that represent the delay associated with a given route. Each tick is considered 1/18 of a second.

LAN segments are typically assigned a value of 1 tick, a T1 link may have a value of 5 to 6 ticks and a 56 Kbps line may have a value of 20 ticks. Larger number of ticks indicate a slower routing path.

Three commonest problems that can occur with RIP are:

  1. Routing loops: the router indicates that the shortest path is back the way the packet came from.
  2. Slow Route Convergence: routers have delay timers that start counting after the RIP advertising packet is broadcasted. This gives the routers time to receive and formulate a proper routing table from the other routers. If the delay timer is too short, the routing table can be implemented with incomplete data causing routing loops
  3. Hop Count Exceeded: the maximum number of hop counts is 15 for RIP. A hop count of 15 is classified as unreachable which makes RIP unsuitable for large networks where hop counts of 15 and above are normal.

EGRP - Exterior Gateway Routing Protocol

EGRP was created to solve many of the problems with RIP and has become the default routing protocol across the Internet. EGRP is an enhanced distance vectoring protocol, it uses up to 5 metrics (conditions) to determine the best route:

  • Bandwidth
  • Hop Count (Delay) - maximum of 255
  • Maximum Packet size
  • Reliability
  • Traffic (Load)

These routing metrics are much more realistic indicators of the best routes compared to simple hop counts.

OSPF - Open Shortest Path First

OSPF is a link state premise, this means that it has several states of routers linked together in a hierarchical routing model:

The top of the root is the Autonomous Router, it connects to other autonomous systems (the Internet). The next is the Backbone Routers, which is the highest area in the OSPF system. Border routers are attached to multiple areas and run multiple copies of the routing algorithm. Last is internal routers which run a single routing database for one area.

Basically, by dividing the network into a routing hierarchy, substantial reduction of routing update traffic and faster route convergence results on a local basis. Each level has a smaller routing table and less to update.

If this page has helped you, please consider donating $1.00 to support the cost of hosting this site, thanks.

Return to

TelecomWorld 101

Copyright July 2013 Eugene Blanchard