Just a basic query, which i would like to get it clarified.
Q1. Why is EIGRP called Distance Vector?
My guess :- Though EIGRP does not use metric as the hop count, it uses delay (and BW)) and delay is the sum of the delay of all the hops to the destination. Each hop is a distance. Is that the reason its called Distance Vector.
Q2. EIGRP routing is called routing by rumour. It get routing updates from its neighbor and belives it that. Compared to this, its said that OSPF has a complete picture of the network and is like a road map. My question is even OSPF gets the information of all the network only from its neighbors? So why this kind of routing update is not called routing by rumour and how is this routing superior to EIGRP routing process.
Thanks in advance.
EIGRP is a bit of a funny one in that it has aspects of both distance vector and link-state. However for your example distance vector does not mean the metric has to be hop count, it can be other metrics. The key thing as you rightly state is that each router receives advertisements from it's neigbor which contain distance values to destination subnets. So it is routing by rumour in that the router is reliant on it's neighbors for accurate distance reports.
OSPF is a link-state. It does not rely on it's neighbors view of the network. It receives LSAs (Link State Advertisements) from neighboring routers or the DR (Designated Router) on ethernet. LSA's are flooded to all routers in an area so every router receives in an area receives exactly the same LSA's ie. no interpretation in terms of hop count etc. is made before the LSA is forwarded on. Only when the router has received all LSA's does it then compute an SPF (Shortest Path First) to all destinations. So each routers calculations are not dependant on any calculations done by their neighbors.
"Compared to this, its said that OSPF has a complete picture of the network and is like a road map."
Jon's post mentions it, but I wanted to emphasis that OSPF doesn't always have a complete picture of the network beyond the local OSPF area.
"EIGRP routing is called routing by rumour."
First time I've heard that. Perhaps all dynamic routing protocols could be considered such?
The whole point of dynamic routing protocols is to convey some type of topology information beyond the next hop. Whatever the information is, it's usually trusted. Which is why some routing protocols also support authentication of the neighbor that provides topology information.
I would suggest not assuming one routing protocol, e.g. EIGRP vs. OSPF, is generally superior to the other. There's a reason there are many. You need to compare the advantages and disadvantages, and how them might impact your network when making such decisions about which routing protocol(s) to use.
"First time I've heard that. Perhaps all dynamic routing protocols could be considered such? "
All distance vector protocols are referred to as routing by rumour. It's not to do with authentication or trust as such it is because each router only sees the rest of the network from the perspective of it's neighbor whereas a link-state protocol such as OSPF does not rely on it's neighbors perspective, rather it computes it's own SPF from the LSA's it receives.
OSPF does have some elements of distance vector in terms of external routes and i perhaps should have mentioned that.
Thank you for the additional information.
Unsurprisingly, confirmed what you describe. I.e., e.g.: "Because each router depends on its neighbors for information, which the neighbors in turn may have learned from their neighbors, and so on, distance vector routing is sometimes facetiously referred to as "routing by rumor."" Still, first time I've recall bumping into this. (Perhaps because I'm not a network engineer - or I keep bad company?)
While researching this, saw most references were to RIP. Makes me wonder whether the term orginally had more to do with static routing vs. early dynamic routing. Reason I wonder is, because distance vector and link state protocols both rely on neighbor(s) to relay information beyond the locally connected interfaces. Both construct their routing table from neighbor provided/relayed information. So, from these aspects both are "routing by rumor". I.e., both rely on relayed information.
Realize there's certainly differences between distance vector and link state, but think just the term "routing by rumor" poorly represents all the differences. (My two bits worth.)
If your primary job is not as a network engineer i'd hate to think how good you are when it comes down to your real job :)
"because distance vector and link state protocols both rely on neighbor(s) to relay information beyond the locally connected interfaces"
Within an OSPF area that is the key. You are not relying on neigbors beyond the locally connected interfaces because on a multi-access network each router forms an adjacency with the DR/BDR and receives LSA's from the DR and the DR is on a local segment. The DR is relaying LSA's but only from other routers that are on the local segment.
Both do construct their routing tables from information received from neigbors but distance vector protocols start with an already calculated route to each subnet from their neighbor whereas link-state protocols don't.
Outside of an area things as you say are slightly different. Perhaps "routing by trust" would be a better term :)
Hi Jon/ Joseph,
Thanks a lot on your time for this discussion. It was really fruitful and helped me clarify a few concepts about OSPF & EIGRP. We tend to overlook these small but very important things.
Thanks again. Cheers!!
I set up a very simple OSPF point-to-point configuration between 2 routers in a lab. Examine the configs and the output of a 'sh ip ospf database router'.
Take note how each router stores LSAs in its database that paint a topological picture that describes the relationship in terms of connectivity between each of the interfaces that is running OSPF. Each router is simply describing the link state for each interface: who is it connected to, what type of network, the LSA header information, etc.
No one is declaring that they know how to get to a certain network and passing it on to a neighbor, and no receiving router is believing them. In reality, all the sending router is doing is recording network information for each directly connected neighbor and passing it on to a receving router that will place the LSAs in a local database. From this, a map of the network is intuitively deduced by the OSPF process.
After having done this, the router will run the so-called SPF (Dijkstra) algorithim to determine which path is shortest to any particular destination network. That path information is what will be entered into the routing table.
It's a complex operation that cannot possibly be described here. You must read it -- and then read it again and again, trust me on this one.
A router running EIGRP, on the other hand, will not advertise the state of a link (A "link" being defined as an interface running EIGRP), but will instead advertise reachability information taken from its routing table. It will inform the EIGRP neighbor that it has a route to a destination network (this is the distance vector behavior) and the receiving router will take note of that information. HOWEVER, because the advertising router does use bandwidth and delay as the default metric to quantify the "cost" to the destination network it is advertising to its neighbor, the receiving EIGRP process can create a topological map that allows it to quantify the "cost" to a destination and then run the DUAL algorithim (Diffused Update Algorithim) to determine the best path to the destination (this is the pseudo link-state behavior of EIGRP).