QoS help..

[lildeezul]

k these are some of the concepts i grabbed by re reading chapter 10 of the BCMSN self study guide.

QOS can use layer 2 and layer 3 classification methods via CoS (class of service) and DSCP,or IP precedence (old) respectively which belong to the ToS byte in a packet. CoS has 3 bits to use for classificatio in the 802.1q frame type. Which gives of a values 0-7.. value 6 and 7 are used for networking protocols such as EIGRP and other features that must have priority , becuase if not there will be no connectivity.value 5 is used for Voice, 4 for Video, 3 -2-1 for data, and 0 for FIFO ( first in first out/ Best effort)

IP-prec uses the 3 most significant bits in the TOS field which gives a value from 0-7.
DSCP ( diff service code points) use the 6 bits which gives 64 values from 0-63, and the last two bits are for ECN..

Classification is the method used to classify packets as a certain type of data, and arrange them into specific groups.. There are four ways to classify data.

1- trust modes via the MLS QOS TRUST DEVICE CISCO-PHONE (cdp used to find phone and trust DSCP values)

2- Manual classification based on DSCP values on incomming traffic ( MLS QOS TRUST DSCP/COS

3- Classification based on class maps, policy maps, and ACL

4- Classification using NBAR ( network based application recognition) THE COOLEST>

Marking is a method to "color" packets/frames so that the next hop doesnt have to deeply inspect the packet to classify the data. Marking makes classification easier becuase all the next hop has to do is look at the tag and say " aww this is voice data, or this is video data"
marking is done via ( MLS QOS DSCP/COS xxxx)

this is what i understand so far.

i am now reading about traffic shaping and policing...

tell me if you think im am getting this ok. ??? there is a lot to learn !

let me know if i have the concepts down

[lildeezul]

more knowlege to me


class maps are used for classification
and policy maps are used to determine what to do with the packet based on the class map
and you can apply policy maps via the SERVICE-POLICY interface config command.


i think im getting it..........but i am getting confused here....

i THOUGHT that classification is based on ingress traffic, but jeremy mentioned that you can use marking on ingress/egress frames..... i though marking would only be on egress frames/packets ( based on Cos or IPpred/dscp) to ease the process for other routers so they do not have to deeply inspect packets.

this above confuses me.. if marking is done on ingress frames then isnt that the same as classifying data ???? that confuses me.

oh wait.. i think i see where we would mark ingress...

lets say for examply off the router port their is a ip phone... if the IPphone is capable of marking thier own data on egress data, via cos 5 ( maps to dscp 40) but the router treats dscp 40 as low traffic priorty, then we would need to mark packets ingress that have a value of CoS 5 (DSCP 40) to a different value so it can use it as a internal DSCP.

would that be an example of this situation of marking ingress ??

[inindix]

Originally Posted by lildeezul

more knowlege to me


class maps are used for classification
and policy maps are used to determine what to do with the packet based on the class map
and you can apply policy maps via the SERVICE-POLICY interface config command.


i think im getting it..........but i am getting confused here....

i THOUGHT that classification is based on ingress traffic, but jeremy mentioned that you can use marking on ingress/egress frames..... i though marking would only be on egress frames/packets ( based on Cos or IPpred/dscp) to ease the process for other routers so they do not have to deeply inspect packets.

this above confuses me.. if marking is done on ingress frames then isnt that the same as classifying data ???? that confuses me.

oh wait.. i think i see where we would mark ingress...

lets say for examply off the router port their is a ip phone... if the IPphone is capable of marking thier own data on egress data, via cos 5 ( maps to dscp 40) but the router treats dscp 40 as low traffic priorty, then we would need to mark packets ingress that have a value of CoS 5 (DSCP 40) to a different value so it can use it as a internal DSCP.

would that be an example of this situation of marking ingress ??

If taking off the router port their is a ip phone and sone IP phone is capable of marking their own data on egress data...




__________________
Call Center Software

[Theacycle]

Is it too much trouble for you to post the Class-map statements the Auto-QoS has generated?Sorry to go on a deviation, but I cant see why the auto-qos would be setting DSCP when the Cisco Phone would have their packets marked by default..Unless Im not seeing something.

[mnop641]

That’s something. Every day a new day, but not for me. ：（