Experts,If I define a Priority percentage for voice traffic and if voice traffic is not using that bandwidth, then can other traffic types can use that unused bandwidth reserved for voice?
as far as I know it still doesn't use in class based weighted fair queuing, even if you use strict prioritization on voice traffic, whatever's leftover been with the iOS automatically re prioritize your other classes and give other traffic bandwidth.
Sent from Yahoo Mail on Android
So other classes of traffic can use the priority bandwidth reserved for voice, if unused by voice?
Simple answer (to my understanding) is yes.
The voice traffic is only prioritized when it's needed. If there are no voice calls on the link, the other traffic can use 100% of the links capacity.
I have one more query.Suppose 1Mb is the priority bandwidth for voice.Currently the whole 1 Mb is unsued by voice. So the other protocol https is using that 1 Mb.Then suddenly voice generates 1 Mb traffic, then what will happen to that 1 Mb of http traffic?
Then suddenly voice generates 1 Mb traffic, then what will happen to that 1 Mb of http traffic?
That's why you have to limit the maximum bandwidth for priority class. Otherwise, it would consume as much as available bandwidth in case of high traffic rates.
QoS is used when you are reaching the defined threshold limits of classification of traffic defined. Remember if you don't reach those tresholds you are not necessarily going to be using the policy in place most of the time. Until limits are exceed to what's configured.
In terms of priority percentage you configure this mainly for voice and video traffic because it has some essential pieces needed in voice. For example : latency / jitter. Voice traffic is sensitive to delay which the priority command helps keep in line for the traffic defined, while the bandwidth command does not. So typically priority is used in the sense of voice and bandwidth in the sense of data.
So lets say I configure the below on a 1 MB link :
policy-map XXX class voice priority percent 10 class data bandwidth percent 30
This does not automatically apply because this is not a guaranteed reservation. We have to hit the thresholds defined for the reservations to be applicable in times of congestion.
So the priority defines a minimum bandwidth guarantee and the bandwidth command provides a guaranteed equal specified rate of bandwidth. The priority queue uses a token
bucket that measures the streams of traffic. This way our voice traffic, for example, is guaranteed low latency and any excess traffic ( policing ) can be dropped in times of congestion.
If you look up "strict priority queue" or "strict de-queuing it will go in further detail of the Layer 3 queues and the buckets based on queueing methods such as LLQ, CBWFQ, WFQ, etc.
I will try to find the reference I used while studying this.So if we are not congested, then yes the bandwidth is shared across the classes including default. However lets say the default
class does not need the traffic but your other classes do. They will use the left over bandwidth percentage as defined in the policy map when configuring " bandwidth percent".
So if I had only 1 MB left over of a 5 MB link and the following config applies:
class data bandwidth percent 30
class critical_data bandwidth percent 50
The two classes will split the traffic based on their defined rates of 50 percent and 30 percent of the 1MB if BOTH need the extra bandwidth.
Hope this helps some.
Assuming voice is configured with 1 mbps priority bandwidth.
If voice traffic throughput is less than 1 mbps, say 768kbps, then the rest of classes can use the remaining 256kbps.
If voice traffic throughput more than 1 mpbs, say 1200kbps, then only 1mbps of voice will pass, and the remaining 200kbps of voice traffic will be dropped.