Can high volume data traffic affect EIGRP performance? e.g. can the traffic make EIGRP flap? My understanding is that the EIGRP traffic must have a high priority, right?
not all Cisco platforms have a system queue for routing protocols for example: on GSRs or C7500 you need to configure a class of traffic for routing protocols to provide them resources when link utilization is high.
On other platforms there is an implicit system queue that by default uses 25% of bandwidth. (this is related to the max-reserved bandwidth command)
<1-100> Max. reservable bandwidth as % of interface bandwidth
In NBMA point to multipoint in addition to provide QoS protection you may need to tune the bandwidth used by EIGRP using
ip bandwidth-percent eigrp 100 ?
<1-999999> Maximum bandwidth percentage that EIGRP may use
this is because if the link is oversubscribed you need to use a low bandwidth parameter.
by default EIGRP shape its traffic at 50% of the bandwidth value
Hope to help
Basically, the traffic itself does NOT affect the EIGRP, I think.
For example, please see the attached document.
After checking it, EIGRP doesn't flap w/o qos.
Also, even if "max-reserved-bandwidth 100" is set, it's the same.
This is because the router internally does the priority processing.
However, when the high CPU utilization by the excess traffic, it could have the EIGRP flap.
But as Giuseppe indicated, 7200 and GSRs behave differently from this.
I am thinking. Whether it is possible that because of the high volume of traffic, the CPU usage in a router can be too high to have time to execute the EIGRP algorithm?
If it is. I wanna know how can you know it is because this high cpu calculation?
>> the CPU usage in a router can be too high to have time to execute the EIGRP algorithm?
This shouldn't happen because the routing processes should be with higher priority.
if EIGRP packets are protected / QoS prioritized on the links by way of implicit or explicit queueing the EIGRP topology should be stable and only hellos are exchanged.
This needs to happen on both sides of each link each router decides how to send traffic and only in some extent can try to provide better treatment to received signalling plane packets.
This a key point because it prevents large topology changes to happen:
if the EIGRP neighborship a new one has to be built and changes are required in EIGRP topology table.
In case of unprotected EIGRP traffic you should see:
the link is empty
the EIGRP neighborship is built
EIGRP updates are exchanged
traffic start to fill the link
EIGRP falls down
traffic goes to zero
An high cpu usage could influence the capacity of the router to answer to a Query in a timely manner.
cpu activity can be monitored using:
sh proc cpu history
sh proc cpu 1min sorted
that lists the processes that use more resources
Hope to help
"Whether it is possible that because of the high volume of traffic, the CPU usage in a router can be too high to have time to execute the EIGRP algorithm?"
I would think its a possibility if the CPU load was at a high substained load. Anything beyond 80% substained average might be cause for concern.
"If it is. I wanna know how can you know it is because this high cpu calculation? "
Giuseppe's post has good suggestions.
What might help (besides reducing the CPU load) if this is in fact an issue, there's config command (two commands that differ, I recall, between IOS versions) to adjust the ratio between interrupt CPU and process CPU. I don't recall the actual commands.
"Can high volume data traffic affect EIGRP performance?"
A typical example is when eigrp routers are so bogged down and congested that they dont respond to a query when a route goes active, thereby leading to an SIA condition.