RIP, IGRP, EIGRP, OSPF comparison

CCNA Level:

Routing Fundamentals:
- Static Routing: Manually update possible routes.
Dynamic Routing: Only need to update adjacent routes.
Default routing:
+ You can only use default routing on stub networks. It is also useful for summarization of static route.
LabC(config)#ip route 192.168.10.0 255.255.255.0 192.168.40.1
LabC(config)#ip route 192.168.20.0 255.255.255.0 192.168.40.1
LabC(config)#ip route 192.168.30.0 255.255.255.0 192.168.40.1

LabC(config)#ip route 0.0.0.0 0.0.0.0 192.168.40.1 (Just one statement can replace all three above)

Where 192.168.40.1 is known as the gateway of last resort.

Three ways of creating default route.
A#ip route 0.0.0.0 0.0.0.0 192.168.40.1 --> A.D = 1
A#ip route 0.0.0.0 0.0.0.0 s0/0 --> A.D = 0
A#ip default-network 192.168.40.0 --> REMEMBER to use network ID!
When IGP(RIP,IGRP,OSPF) is configured, default-network would be advertised.

Thus, A.D of static route by default = 1 (unless you wish to change it)
Connected interface = 0
static route = 1
EIGRP = 90
IGRP = 100
OSPF = 110
RIP = 120
External EIGRP (AS redistribution) = 170
Unknown =255 (never be used)

- When routing between routers, layer 3 or above is used.
When the transferring packets between interfaces, layer 2 or below is used.
- Routing from host to router.
ICMP(Layer 3) à IP (Layer 3) à Packet (layer 3) --> ARP (Layer 3) à Frame-Encapsulation(layer 2) à FCS-CRC (layer 2) à Packet (layer 3).
Router will then do the same thing by pushing the packet into the next interface
- Rule of thumb:
The destination MAC address will always be the router's interface (if you packets are destined for outside the LAN), never hub/bridge/switches.

EIGRP
- Dynamic routing (support very large network)
- classless protocol (subnet mask is preferred to be specified during IP addressing)
+ VLSM (Variable length subnet mask) / CIDR, not by default - seldom asked in the exams
+ Give rise to discontiguous network
+ Auto summarization (forced summarization)
--> Therefore, when advertised for network, you NO need to provide the mask information.
- concept of Autonomous Aystem (share routing information)
+ neighbor discovery
+ Mechanism:
+ Hello or ACK received
+ AS numbers match
+ Identical metrics (K values)
+ This will create the topology table.
- used only when all routers are Cisco
- hybrid : send distance-vector and link-state packets
+ distance-vector
+ Feasible distance = reported distance + metric to the neighbor reporting the route
+ Reported distance =
- Event-driven (hold-down value = 0)
+ suitable for very large networks. (maximum hop count of 255)
- Supported all routed protocols (IP, IPX, AppleTalk)
+ Due to its protocol-dependent modules (PDMs)
- Communication via RTP (Reliable Transport Protocol)
- Reliability is the focus
- uses multi-cast and unicast (when no reply is received from particular router, 16 times max a.k.a reliable multicast).
- Best path selection via DUAL (Diffusing update Algorithm)
+ Topology table is used as extension to the routing table, for best path analysis. Means tat will be used as input towards DUAL alogorithm
+ Fastest routing protocol (a.k.a fastest route convergence )
+ Each routers having a copy of neighbors' routes (neighbor table)
+ Keep updating the neighbor table
- Load balance of up to 6 links (based on metrics)
+ Metrics is refer to link cost (which is then bandwidth and delay by default), however, EIGRP may use a combination of four.(bandwidth, delay, load, reliability)
- Support "passive interfaces", similar to switch "port-fast"

Others:
- EIGRP routers that belong to different A.S can't share information automatically (they don't become neighbors). This is a good practice to reduce bandwidth.
If you still insist to share information between these routers, then you have to do it manually. This is known as redistribution.
- Internal EIGRP A.D = 90
External EIGRP A.D = 170 (when redistribution[automatic or manual] happens)
+ Automatically redistribution - In an existing IGRP network, putting a router with EIGRP having the same A.S as the existing IGRP will bring about automatic redistribution.

- A feasible successor is a backup route and is stored only in the topoloyg table. A successor route is stored in the topology and also the routing table.
Having feasible successor helps with network convergence (thus no need to study the network anymore, just need to update changes)

OSFP:
- Dynamic Routing (large network support)
- classless protocol (subnet mask is preferred to be specified during IP addressing)
+ VLSM/CIDR (not by default)
+ Gives rise to discontiguous network, therefore catering for huge network (reduce bandwidth)
+ Route redistribution - a translation service between routing protocols.
+ No auto-summarization BUT MANUAL summarization.
--> Therefore, when advertised for network, you have to specify the reversed mask. However, you if don't specify the mask information, then it will assume as the default mask.
- Send link-state packets
- Supported only IP routed protocol.
- Minimize routing update traffic (the one with lowest management bandwidth requirement)
--> Because it is event triggered
- Hierarchical network. Because it is using Area (even though everybody has to connect to Area 0). "Area" is actually a subset for Autonomous system (in EIGRP). Typically, OSPF runs inside an autonomous system.
- Topology database is also known as "link-state database". It is used as input to the algorithm for shortest path computation.
- OSPF can work with layer 2 networks of broadcast-multi-access(i.e Ethernet) and Non-broadcast-multi-access(NBMA) (Frame Relay, X.25, ATM), as well as point to point and point to multipoint.
BMA, Point to mulitpoint = DR and BDR needed
NBMA, point to point = DR and BDR not needed

- OSPF process ID:
+ Locally significant only
+ Grouping of commands under specific running process. (Each process maintain separate copy of its topology table)
+ Enable OSPF routing on the router.
- Recommended to use loopback interfaces because it will always be active, and since loopback interfaces' IP will become the router ID, this will also affect selection for DR and BDR.

RIPv2:
- classless protocol
- Auto-summarization (forced summarization)
--> Therefore, when advertised for network, NO need to specify the netmask.
- Supported only IP routed protocol.
- Distance-vector protocol
- Gives rise to pinhole congestion
- Slow convergence time.
+ Gives rise to inconsistent routing tables and routing loops (when update not done simultaneously).
Routing loop counter mechanism:
+ Split horizon = never send back to same direction
+ Route poisoning = set the hop count in routing table to 16

Routing loop counter tool:
+ Holddowns timer
+ Maximum hop count =15 (the number 15 is considered as a loop)
--> Split horizon and route poisoning rely on holddowns timer and maximum hop count for its working. --> Exam question
+ Not suitable for large network.

RIPv1:
- classfull protocol
+ No VLSM
- Auto-summarization (forced summarization)
--> Therefore, when advertised for network, NO need to specify the netmask.
- Supported only IP routed protocol
- Load balance of up to four links (with equal cost links, that is hop count.)
- Distance-vector protocol
- Gives rise to pinhole congestion
- Slow convergence time.
+ Gives rise to inconsistent routing tables and routing loops (when update not done simultaneously).
Routing loop counter mechanism:
+ Split horizon = never send back to same direction
+ Route poisoning = set the hop count in routing table to 16

Routing loop counter tool:
+ Holddowns timer
+ Maximum hop count =15 (the number 15 is considered as a loop)
--> Split horizon and route poisoning rely on holddowns timer and maximum hop count for its working. --> Exam question
+ Not suitable for large network.

IGRP
- classfull protocol
+ No VLSM
- Supported only IP routed protocol
- Distance-vector protocol
- Maximum hop count of 255, however uses bandwidth and delay as the metric instead.

All in one:
- RIPv2 and EIGRP support discontiguous networking, but no by default. OSPF does support discontiguous networking by default because it does not auto summarize classful boundaries as RIP and IGRP do.
-By default, RIP can load balance across four links as well. However, unlike IGRP and EIGRP, they must be equal links (hop count)
-Under RIP, the passive-interface command will prohibit the sending of route updates but allow their receipt. Thus, a RIP router with a passive interface will still learn about the networks advertised by other routers. This is different from EIGRP, where a passive-interface will neither send nor receive updates.
- Multi vendors environment --> RIPv1 (outdated), RIPv2 (smaller network), OSPF (large network)
Cisco only --> IGRP(outdated), EIGRP (large network)
- Static route between two ends has the less bandwidth requirement.
- Link-state protocol has more intelligence than distance-vector protocol. Hybrid is the king.
- RIP & IGRP can only be used on a classful network (with consistent subnet masks)

Tips:
- Always compare OSPF to RIPv2 and RIPv1 because these are open-standard. Thus, within an all cisco-router environment, EIGRP would be the best choice.
- Due to slow convergence nature, RIPv1 and RIPv2 requires too much bandwidth, thus OSPF is preferred.
- classfull protocol means either:
1. Don't think about subnet / ignore subnet
2. Use the most significant octets

For instance: the classful of 172.16.0.0 /24 is 172.16.0.0
not 172.16.10.0

Comments

usaf said…
In the very last line i think you meant this:

For instance: the classful of 172.16.10.0 /24 is 172.16.0.0
not 172.16.10.0
Anonymous said…
Very Interesting...
ibraheem said…
I looked the previews of the site i really liked it because it has a so many better topic but liked the it topic.For more info:
click here for online degrees .
smrutiranjan said…
good job............