I am having issues configuring QoS from start to end on ArubaOS 7.4.1, this is a broad topic, with many different type of answers.
I understand that I need to configure things in this order
Classifier
Policer
Queueing
Classfier: using a 1120E Nortel
Let's start with classifying, let's assume I want to use LLDP-MED (Media Endpoint Device) to have my phones mark packets. All I need to do then is enable 802.1ab on the phone, is this correct?
Then create two VLANs, one for Data, and one for Voice, using the
#interface-profile voip-profile
to distinguish the Voice Vlan, of course applying it to the access port.
Is this correct? Now what about the Data VLAN, how can I point the phone to the Data Vlan, what command do I use? Is this an interface-profile?
To further build on the voip-profile config,
#voip-dot1p
This is whatever or however I want to mark my ToS byte, I am assuming preffered 6 for voice. Is this correct?
#voip-dscp
This is my DSCP field, and used to mark traffic as well, this could be any number I wish for my Voice traffic, but I am assuming preffered is 46, is this a correct assumption?
#voip-mode auto-discover
I am assuming I would use Auto-Discover because I enabled LLDP 802.1ab on my phone, which also enabled (I hope) LLDPMED as well on the 1120E Nortel handset. Is this a correct configuration, and assumption?
Now that my voice traffic is marked, or classified, how do I now make sure that Data traffic is in the correct VLAN and also marked in the ToS field? Is there a specific configuration I am missing?
Apply a Policy,
I now need to use the
#policer-profile (name)
command to build a poilcy on what to do with packets as they pass through. I am assuming this will direct the traffic into specific queues for processing. Like Voice traffic marked ToS 6 with DSCP of 46 will eventually be put into a Low Latency Queue for priority processing above all TCP traffic. Is this a correct assumption?
#Qos-profile Voice
BEFORE I can write the policer profile, however, I need to write a qos-profile. Is this a correction assumption?
So I now write a Voice qos-profile, I assume, to find, and direct, specifically marked packets using the following:
#qos-profile Voice
#dot1p 6
Does #dot1p 6 "re-write" over whatever the ToS field was before?
Is this not a 'classfier' function?
How does this command work, why do I use it?
#dscp 46
Does #DSCP 46 "re-write" over whatever the DSCP field was before? Is this not a 'classfier' function?
Now what do I do with the following?
(Please if you could explain why as well if possible,)
#drop-precedence
#traffic-class
Is this function used to mark into one of the 7 available queues on an interface? How do I effectively use this command?
Policer-profile cont'd
I now write a policer-profile Voice, and apply it to an interface, lets say the access port for the phone.
Now I have my qos-profile Voice ready to go, where do I apply it to the policer? How do I properly configure a Policer?
Is this my Queueing tool? Is this one of the 7 queues that are available per an interface?
How do I use this properly to create a Low Latency Queue for Voice packets?
How do I use this to create a Class Based Fair Weighted Queue, for all other Data traffic?
What the heck is
#violate-action remark
Remark what? Where? In a ToS field?
I feel like I have 33% of this understood maybe a bit more.
How do I manage my queue and create a LLQ-CBWFQ, for both voice and all other data?
Thank you very much in advance, I know this is a whole lot to sift through.