Rate Limit egress data on multiple VLANS trunked over 1 port.

[Lee Smitherman]

Hi,

I have a sync link which is a Gb L2 p2p circuit between datacenters.  The A end connects to a 3850 802.1q trunked port and the B end connects to a Nexus 5k 802.1q trunked port.  We simply have vlans (non svi, just L2 backend vlans) traversing the link for each application that uses this link for data synchronisation.  Question is, out of that 1 Gb, how did I give for example Vlan 1 a rate limit of 200MB, Vlan 2 a rate limit of 100Mb etc.   Im guessing I do need to police here, however its above my knowledge of Qos.

Lee.

[Zach S · Accepted solution]

Lee,

If all you are trying to do is cut off VLANs at specific rates like that, the QoS implementation looks like it would be fairly easy. The real difficulty is choosing the correct policies that will still allow your services to work correctly within those limitations. Such as, if you stick with a single lump policing policy for VLAN1, then all the traffic will be treated the same and UDP traffic will be dropped during peaks with no preference over TCP; so if you had any VoIP or other sensitive UDP traffic it would suffer significantly, additionally you run the risk of TCP sessions timing out if too much traffic policing happens.

You can match based on VLAN tag and apply an action, such as shaping or policing. The big difference being that shaping buffers excess traffic, whereas policing drops it flat out; there are other options and considerations as well, which makes QoS the difficult subject that it is.

More info: http://www.cisco.com/c/en/us/td/docs/ios-xml/ios/qos_classn/configuration/xe-3s/qos-classn-xe-3s-book/qos-classn-vlan.html

Shaping vs. Policing: http://www.cisco.com/c/en/us/support/docs/quality-of-service-qos/qos-policing/19645-policevsshape.html