Here is what we try to do:
We have an S6 installed in our data center, and we have several customer sites connected to dedicated ports. Most are 1 single uplink ports, some are lags over 2 ports for redundancy.
Due to contracts with our customers, the contractual bandwidth is sometimes lower than the physical link provided via MPLS/VPN/whatever. For example, we have one customer that is connected with a raw fiber links and thus have a physical connection of 1G, but they only pay for 100Mbps...
Therefore we have a demand to apply traffic shaping on both inbound and outbound traffic per port / lag:
- Customer A: port 5.1, 100Mbps
- Customer B: port 5.2;6.2, 200Mbps
- ...
I have tried this a few time by using Policies and COS (via Netsight), but never with full success. This morning we tried to limit one 1G uplink to 100Mbps, but that changed speed tests (www.speedtest.net) from 350/900 (download/upload) to 350/60 Mbps:
- Download not effected at all
- Uplink reduced much more than intended - unless this is expected with speedtests over HTTP?
I tried to follow this guideline:
https://extremeportal.force.com/ExtrArticleDetail?an=000078687
I implemented this, except for the port group... The COS profile looks like this:
Which opens another question: Are these COS profiles shared? Meaning: If I assign the 100Mbps profile to 2 ports (= 2 customers), are they sharing 100Mbps, or do they get 100 each? Or do I have to create a dedicated COS profile per customer/port?
