Tx and Rx Load not increasing when pinging

[tiwariharish44]

Hi,

We have 10Gbps connectivity between distribution(4507R+E) and Core(Nexus 7010). I want to ask why the Tx & Rx load is always 1/255. Even when I ping the routed interface with a large ICMP data size, still the load does not increase. We are using EIGRP between Core and Distribution

Thanks

[Peter Paluch]

Hello,

I do not believe that you can create any significant load on a 10Gbps link by sending a series of pings. To my best observation, subsequent pings are sent at most at a rate at which responses are received back, and in network operating systems, replying to pings is a relatively low-priority process. It is relatively normal to receive a ping back in 1ms, so the pings can be realistically sent at a rate of 1 ping per 1ms, or 1000 pings in 1 second. Even if we used max-sized pings which would be roughly 9 kilobytes per ping (give or take a few hundred bytes), the total rate would be approximately 1000 * 9000 * 8 = 72000000 bits per second, or 72 Mbits per second. This is a mere 0.72% of the capacity of a 10Gbps link. As the Tx/Rx load is expressed in fractions of 1/255, 1/255 is equal to 10Gbps/255 = 39.22 Mbps. So, assuming that you have been pinging with 9000-byte pings, and that they were sent at a rate of 1000 pings in a second, the incurred Tx load could have been 2/255 at most.

So I am not surprised at your findings - I'd say they're normal.

Best regards,
Peter

[Steve Fuller]

Hi,

I think it’s simply that the ping is not generating sufficient traffic.

For a 10 gigabit Ethernet link, each 255th equates to 39,215,686 bit per second, or 4,901,961 bytes per second. If you’re sending 1400 byte packets you would need to generate 3,501 packets per second for each 255th.

The router simply doesn’t generate pings that fast.

Regards

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

To add another point, L3 switches process data plane traffic with dedicated hardware, and control plane traffic (generally) with a general CPU.  As on these devices most of the "heavy lifting" (the data plane) is processed by dedicated hardware, the general CPU is often on the "small" side.  Ping processing is generally dealt with as control plane processing.  So, even if the L3 switch's software wanted to generate pings at full 10g line rate, there's a good chance the CPU would be unable to support it.