Interior versus Exterior
- Used within a common administrative domain called an Autonomous System (AS), which is typically controlled by a single authority or company
- Interior routing protocols are used within a corporate network
- Used to connect Autonomous Systems
- Exchanges routing information between different administrative domains
- Exterior protocols are used to connect sites within a very large corporate network, or are used to connect to the Internet
Distance Vector versus Link State versus Advanced Distance Vector
- Maintains a routing table of distance and vector (direction) to each network.
- Sends periodic routing updates that include the “entire” routing table at each update, to its neighbors.
- Received routing updates are processed, and the resulting routing table is sent by the router to its neighbors. Thus the updates are “second-hand information” (routing by rumor)
- Prone to routing loops (disagreement between routers) and count to infinity (routing metrics continue to accumulate indefinitely)
- Solutions to these problems include:
- Split Horizon: do not send updates back out the interface through which they were learned; this eliminates back-to-back router loops
- Maximum metric: eliminates count to infinity problem
- Route poisoning: set the advertised metric to the maximum value on routes that have gone down
- Poison reverse: overrides split horizon by informing the source of a route that it has gone down
- Hold-down timers: eliminates long-distance loops by ignoring updates about “possibly down” routes that have metrics worse than the current metric
- Triggered updates: send an individual update immediately when a route is thought to be down, rather than wait for the periodic update timer (also called flash updates)
- Maintains a complete topological map (database) of the entire network, separate from the routing table (forwarding table)
- Sends updates only when necessary and only sends information that has changed, not the entire database. Does not send information from the routing table, but rather from the database.
- The initial routing update is sent to every link state router in the network (flooding) via a multicast IP address, not a processed copy as with distance vector protocols
- Routing table is individually calculated on each router from its database. This process is called Shortest Path First (SPF). When SPF runs, it is CPU intensive.
- The database typically requires as much memory as the routing table
- Uses “hello” packets to maintain a database of link state neighbors throughout the network
Advanced Distance Vector
- Uses “hello” packets to maintain a neighbor table
- Sends updates only when necessary; only sends information that has changed
- Maintains topology table containing all of the routing information received from its neighbors
- The neighbor’s metric is the advertised distance (AD). The AD is added to the metric to get to the neighbor; this sum is the feasible distance (FD)
- Uses the Diffusing Update Algorithm (DUAL) to calculate best paths. The route with the lowest FD is the best path; it is called the “current successor” route and is put in the routing table. The FD of the current successor becomes the metric in the routing table
- Non-best routes that pass the “feasibility condition” are called feasible successors and can be used if the best route goes away. Feasible successors are kept in the topology table. The “feasibility condition” is: A route is a feasible successor if its AD is less than the FD of the current successor
Classless versus Classful
- Sends subnet mask with routing updates
- Supports VLSM, discontiguous networks, and does not have to automatically summarize routes at a major network boundary
- Does not send subnet mask with routing updates
- Does not support VLSM, discontiguous networks; automatically summarizes routes at a major network boundary
Router Boot Sequence
- Power on self test (POST)
- Load and run bootstrap code
- Find IOS:
- Check boot field of configuration register; if it is 0x2 through 0xF, look in NVRAM for “boot system” commands. Do what they say if there are any. If boot field is another value, do what it says to do.
- Otherwise default to use first valid IOS file in flash.
- If that fails, attempt boot from network server and then from helper image in ROM (if there is one).
- If those fail, use ROM Monitor mode.
- Load IOS into RAM
- Find configuration, if it exists
- Load configuration into RAM
Excerpted and available for download from Global Knowledge White Paper: CCNA v1.1 Exam Review: Critical Concepts of the 640 – 802 CCNA Exam