QOS shaping and policing

hedhrts

Have concepts down, but having trouble wrapping my head around the details. This is probably due to inconsistant units. One is measured in bytes (policying) and the other in bits (shaping). Does anyone have any suggestions on info on the web that could make this easier to nail down, or do I need to keep banging my head against the wall until I get through?

Goldmember

http://www.cisco.com/en/US/docs/ios/12_4/qos/configuration/guide/qchpolsh_ps6350_TSD_Products_Configuration_Guide_Chapter.html


Regulating the packet flow (that is, the flow of traffic) on the network is also known as traffic shaping. Traffic shaping allows you to control the speed of traffic leaving an interface. This way, you can match the flow of the traffic to the speed of the interface receiving the packet.

Poison Reverse

qos might be the most boring thing i've encountered on my cisco studies.

I just kept re reading until it got to me

tech-airman

hedhrts wrote:

Have concepts down, but having trouble wrapping my head around the details. This is probably due to inconsistant units. One is measured in bytes (policying) and the other in bits (shaping). Does anyone have any suggestions on info on the web that could make this easier to nail down, or do I need to keep banging my head against the wall until I get through?

hedhrts,

I too was having difficulty understanding QoS when I was trying to study for the 640-861 DESGN exam. Generally speaking, the way I remembered it is, policying deals with INBOUND traffic and shaping deals with OUTGOING traffic.

The way I remember "policying" is to imagine a police officer standing outside the ENTRANCE door, say, a night club. Imagine the ropes holding the line in as the "input buffer" for the networking device interface. So the various methods of policying is what that police officer is supposed to deal with the line if it gets too long.

The way I remember "shaping" is to imagine a stick of clay that's wide. Well, due to the bandwidth of the outbound interface, say a serial interface set to 1.544 Mbps, that wide stick of clay needs to be "shaped" from it's current "full bandwidth" to a "narrower bandwidth" so it'll fit the outgoing interface. That's why methods like "compression" is considered a "shaping" method.

Now, your confusion between policying units and shaping units is because of various processing that can occur between the inbound interface and the outbound interface. For example, from an inbound Ethernet interface of 10 Mbps to an outbound Serial interface of 1.544 Mbps. Now, imagine a host on the Ethernet network trying to send an e-mail out of the Serial Internet interface. Now, anyone who's ever used a small funnel will know that if you fill up the funnel too fast, the funnel overflows. So to avoid overflowing, you slow down the input of the fluid into the funnel. In other words, you're policing the inbound fluid by slowing down the input rate. Going back to networking, since most literature about the data payload, the TCP/UDP header, the IP header, and even the ethernet header and trailer uses the units of bytes, it is reasonable to understand why policing units are in bytes.

As previously mentioned, compression is a shaping method to make more efficient use of the relatively slower outbound interface, in this case Serial. Since compression is a method of reducing the size of the inbound frame to a smaller outbound frame, maintaining the use of bytes as units is going to be difficult. So just like with mathematical fractions it helps to find the "lowest common denominator," networkingly the "lowest common denominator" is the bit and not the byte.

Here's an introductory overview of Quality of Service from Cisco's website.

Link:

1. Quality of Service(QoS) - http://www.cisco.com/univercd/cc/td/doc/cisintwk/ito_doc/qos.htm