What happens every 60 seconds on a port with rate-limiting configured?

[Paul_Thornton]

I've been doing some testing with rate limiting on an X440 in the lab. Fairly standard, boring, test setup.

Ingress using a meter:
configure meter rl_50meg committed-rate 50000 Kbps max-burst-size 50000 Kb out-actions drop

Egress using rate-limit egress:
config port 3 rate-limit egress 50000 Kbps max-burst-size 50000 Kb

The policy applied to ingress on port 3 just sets the meter rl_50meg, nothing else.

The rate limit does as you expect, and graphing at 1 minute intervals whilst pushing traffic through gives a fairly straight line at around 49Mbit/sec. All happy there.

When the sample period is decreased (ie: resolution increases) some strange artefacts start being observed. At a 5 second poll interval (I didn't go any faster) there are distinct spikes in the graph up to 60Mbit every 60 seconds (see graph below). Is this related to the way the hardware is performing the rate limiting?

This is more a "Curious to know what's happening" question than a problem.

Thanks

Paul.

[eyeV]

Hello. Maybe this post would be useful https://community.extremenetworks.com/extreme/topics/rate-limit?utm_source=notification&utm_medi...

Hi everybody.
It might be interesting for somebody. I've got some explanations from TAC. My problem with rate-limit is peculiarity or feature of platform.

One second is divided into 15.625 microseconds intervals. The rate-limiting mechanism occurs when the platform receives lots of packets in one 15.625 microsecond interval.

For example. I've configured
conf ports 25 rate-limit flood broadcast 100000

Rate-limiting mechanism occurs when the box receives ~1.5 packet in 15.625 microsecond.

[EtherMan-Dupe]

wonder if what you are seeing is due to how often the switch polls and gathers information itself.... so if it cant poll ports stats faster than say 10 seconds would this skew your polling and graphing .... Also is the graph coming from the switch interface or from you test gear ? ... I am also curious about how the buffer size affects the rate shape and where that sweat spot is ....