06-08-2014 04:46 PM - edited 03-04-2019 11:07 PM
Hi everybody
I hope you guys are doing fine.
http://www.cisco.com/c/en/us/support/do
cs/ip/border-gateway-protocol-bgp/107615-highcpu-bgp.html#understandbgp
From the above link:
While BGP scanner runs, low priority processes need to wait a longer time to access the CPU. One low priority process controls Internet Control Message Protocol (ICMP) packets such as pings. Packets destined to or originated from the router may experience higher than expected latency since the ICMP process must wait behind BGP scanner. The cycle is that BGP scanner runs for some time and suspends itself, and then ICMP runs. In contrast, pings sent through a router should be switched via Cisco Express Forwarding (CEF) and should not experience any additional latency.
1)Are we saying the pings sourced from the router will be switched using CEF above? Just want to confirm that.
Please consider the following example:
CE----------R1-----ISP CORE---
Above CE is customer router connected to R1 for internet connectivity. R1 carries Internet routes in a service provider network
So If we issue pings sourced from R1 to CE, will this pings be CEF switched? will the ping replies from CE destined to R1,given lowest ICMP priority, even though CE is responding to R1 pings?
Thanks
Solved! Go to Solution.
06-09-2014 01:39 AM
Hello Sarah,
I hope you are well
ICMP packets originated by the router or destinated to the router are processed by main cpu not by CEF, they are actually placed in an internal queue waiting to be process switched by the main cpu and this may expose them to some delay as explained in the link about BGP scanner. (this is called punting to the main cpu the packets are punted). The same happens for protocol messsages.
The reason is that the packet needs to be examined to prepare the appropriate answer.
So going to your example, the CE node will process switch the ICMP echo request coming from R1 and will process switch the ICMP echo reply to R1, the same happens for R1, that has a full internet table, it process switches its own requests and CE's replies.
So it is R1 the more likely device to introduce delay variation.
in other words the RTT of ICMP packets may vary depending on when they are issued.
Traffic going through the router is CEF switched and far less impacted by BGP scan process.
Hope to help
Giuseppe
.
06-09-2014 01:39 AM
Hello Sarah,
I hope you are well
ICMP packets originated by the router or destinated to the router are processed by main cpu not by CEF, they are actually placed in an internal queue waiting to be process switched by the main cpu and this may expose them to some delay as explained in the link about BGP scanner. (this is called punting to the main cpu the packets are punted). The same happens for protocol messsages.
The reason is that the packet needs to be examined to prepare the appropriate answer.
So going to your example, the CE node will process switch the ICMP echo request coming from R1 and will process switch the ICMP echo reply to R1, the same happens for R1, that has a full internet table, it process switches its own requests and CE's replies.
So it is R1 the more likely device to introduce delay variation.
in other words the RTT of ICMP packets may vary depending on when they are issued.
Traffic going through the router is CEF switched and far less impacted by BGP scan process.
Hope to help
Giuseppe
.
06-09-2014 09:26 PM
Hi Giuseppe
long time no see, thanks for as usual awesome reply.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide