EIGRP uses Least Bandwidth!! isn't that link-state behavior ?

[mohammed hashim]

hi,

EIGRP calculate the bandwidth as:

Bandwidth = (10⁷ / Least bandwidth in kilobits per second)

Least bandwidth means the lowest bandwidth link to the final destination !!

is not that link-state behavior ? 

[Mark Malone]

Hey

EIGRP is hybrid has both distance vector and link state properties in it.

saying that the calculations that EIGRP use are independent to itself ,it uses the K1 metrics to decide which is the best path choice and paths can be manipulated by altering the metrics whether bw/delay etc  ,classic 32bit EIGRP mode just bandwidth and delay If I remember right by default but the new named 64 bit uses all metrics for more precise calculations called wide metrics. This makes EIGRP very flexible in real world and the fact it only exchanges changes that occur limits the resources required on routers when something is dropped unlike link state that will pass the whole table over

[Giuseppe Larosa]

Hello Mohammed,

default EIGRP metric calculation is based on delay and bandwidth.

It is cumulative on delay (sum of all delays of links on the whole path in tens of microseconds multiplied by 256) and it is inversely proportional to minimum BW in the path multiplied 256.

As Mark has noted EIGRP has some characteristics of link state protocols and some of distance vector protocols and for this reason is often classified as a hybrid protocol.

Recent IOS images have added support for 64 bit metric that is useful for links with speed higher then 10 Gbps, however before adding the route to the IP routing table the metric is scaled down by a scaling factor (its default value is 128) because metric field in the IP routing table is still a 32 bit unsigned integer.

EIGRP has the peculiarity that when one router looks for an alternate route it starts to ask to all of its neighbors using EIGRP queries and expect answers from all of them.

This is totally different from OSPF and ISIS where each router runs the SPF algorithm by itself using as input data the link state data base that is guaranteed to have the same contents on routers in the same area ( OSPF case) or same level ( ISIS case).

Link state protocol resend routing information at regular intervals, each data structure contains an age and before max age expires (an hour for OSPF) the data structure originator node sends a new version of the data structure.

The total bandwidth consumed by these updates averaged over one hour is not very big.

Link state protocols are preferred in service provider environments, because they support MPLS traffic engineering extensions, that EIGRP does not support.

In addition to this, EIGRP is proprietary even if recently has became public domain.

https://tools.ietf.org/html/draft-savage-eigrp-00

Hope to help

Giuseppe

[Rolf Fischer]

Hi,

one more little note in addition to the excellent replys you already got.

Least bandwidth means the lowest bandwidth link to the final destination !!

Is not that link-state behavior ? 

Bandwidth-related metrics is not really a characteristic of link state protocols. The standards of OSPF and IS-IS both define cost as a dimensionless metric. The Cisco OSPF implementation (and others) use the interface bandwidth to calculate cost but there are other vendors that set the cost of the OSPF interfaces to 1 by default (hop count). And the default cost for IS-IS interfaces is 10 in the Cisco implementation.

HTH

Rolf

[Jon Marshall]

EIGRP is an advanced distance vector routing protocol.

By default it uses bandwidth and delay to calculate the metric of a route but this does not make it a link state protocol.

With a link state protocol such as OSPF within an area all routers exchange information (LSAs) about their individual links which means every router has a full picture of the area.

Using that database each router can then run the SPF algorithm to calculate the best path to every destination within that area.

EIGRP does not do this.

What EIGRP does is to send routing updates containing the individual metrics such as bandwidth, delay etc. and then each router makes it's own calculation based on the information it received from it's neighbor.

It is still relying on information from it's neighbor's point of view ie. it is not receiving information about every other routers links,  and that is why it is still a distance vector protocol rather than a link state protocol, it's just that it is using more complicated metrics than for example, RIP, which simply uses hop count.

Note also that even with OSPF it is only link state within an area, between areas it becomes distance vector.

Jon