basic qos idea

nobish

Hi,

I am a beginner on Qos. Waned to know about a basic stuff from you. My reading has been stuck on these topic.

Consider a simple scenerio like this:

Internet cloud

---

> (Fa0/1)layer 3 switch (all other Fa ports)

---

>all client workstation.

Please help me clearing the following basics:

1. Here, where I should implement the QoS. Seems like I should do it on the L3 fa0/1 port. Generally, most of the traffic are coming in to the switch's Fa0/1 port as incoming. Also, al traffic going out with the same port. However, I see most cases the policy is implemented as "ouput". Could you plz give me a very brief idea about this logic.

2. I wanna be clear about ingress and egress. Suppose one mail is coming from the internet to a worstation attached with Fa0/2. So, can I get a brief about ingress and egress regarding this scenerio.

GT-Rob

Well the connection to the internet is probably going to be much slower than the port speed of your fa0/1. So what happens is say your internet is 3mb, and that port is trying to send 50mb of traffic from those hosts, 47mbs is just going to get dropped.

QoS doesn't solve the problem of that traffic getting dropped, what it does it let you choose WHAT gets dropped.


So you could put a policy on the output queue of the fa0/1, so say what gets through in that 3mb, and what gets dropped (if its going over).



Remember that the problem isn't that too much traffic is reaching your switch (input), its that fact that it can't send it all (output), which is why the policy usually goes on the output.