Cisco Support Community
Community Member

asr9k inline netflow

so, what truly does it mean in a lot of our documentation slides (networkers, courseware, etc.) that describe the functionality of the asics and cpu on an asr9k line card.      specifically, in the 'box' containing cpu tasks, functionality it lists   'Inline NetFlow"    with no supporting information.

does this mean we perform netflow at line rate, via the cpu?    that makes my head hurt.      netflow is  'process switched' at line rate?

i've found some interesting items so far...

"NetFlow consumes hardware resources (ternary content-addressable memory [TCAM], CPU, etc.), so understanding the resource utilization on a device is important before enabling NetFlow."     from the cisco wiki   but this was in reference to comparing netflow in nx v ios

"Distributed inline Netflowà1M entries/LC , 100Kpps /LC, flexible sampling and IPv4/IPv6/MPLS feature"    in a networkers preso

"In this paper, we propose an online statistical traffic classifier using the C4.5 machine learning algorithm running on the NetFPGA platform. Our NetFPGA classifier is constructed by adding three main modules to the NetFPGA reference switch design; a Netflow module, a feature extractor module, and a C4.5 search tree classifier. The proposed classifier is able to classify the input traffics at the maximum line speed of the NetFPGA platform, i.e. 8 Gbps without any packet loss. Our method is based on the statistical features of the first few packets of a flow. The flow is classified just a few micro seconds after receiving the desired number of packets."      from a scientific paper proposal.


can anyone clarify or distill this into 1 simple answer?


thanks in advance.  


Everyone's tags (1)
Community Member

For Trident HW this is 100

For Trident HW this is 100 kpps per LC, for Typhoon it is 200 kpps per LC. So you need to divide this rate by the number of NPs you use on the LC, e.g. on a MOD160 you have 4 NPUs, so you can do 50 kpps per NPU. As these numbers are quite low it is highly recommended to use sampling for netflow on the ASR9K.



Community Member

thank you Florian!

thank you Florian!

Cisco Employee

TR,note that the netflow


note that the netflow implementation in the a9k is not HW or "inline". Basically when a packet is subject to netflow a "copy" is created and punted to the LC CPU where it is processed for netflow.

More detail is available here:



Xander Thuijs CCIE #6775 Principal Engineer ASR9000, CRS, NCS6000 & IOS-XR
Community Member

thank you Xander!   :)   

thank you Xander!   :)  


CreatePlease to create content