Re: question in voice QOS about queue & gurantee bw
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
There's (hardware) interrupts whether traffic is in LLQ or non-LLQ class. Both should be transparent to your download.
What actually happens on most routers (since you've asked about LLQ), egress traffic attempts to be sent to an interface hardware queue (tx-ring). If that (FIFO) queue/buffer fills, excess packets are stored elsewhere. If you've using CBWFQ, the packets are stored into the queues defined by your CBWFQ policy.
If there are packets in the CBWFQ queues, when the hardware queue has new room (i.e. some packets have been transmitted), software dequeues packets from the CBWFQ queues based on the CBWFQ policy and based on previous dequeuing history. LLQ, if any pakets are queued, is dequeued from its queue first. Non-LLQ queues are proportionally dequeued. So, for example, if there's two queues with packets, but one has twice the bandwidth allocation, the dequeuing will attempt to dequeue twice as many bits from the its class queue relative to the other class queue.
As Rolf noted, LLQ has priority over all other classes, when the CBWFQ attempts to dequeue from the software classes. This to minimize any possible latency and/or jitter. (NB: you can still have latency and jitter even with LLQ.) BTW, you can have multiple LLQ classes, but there's only one actual LLQ (excluding perhaps ASRs).
One other important difference between LLQ and non-LLQ, the former will police LLQ queued packets. The latter will allow additional bandwidth to be used beyond the class allocations. (NB: if packets are not being queued to software queues, LLQ policer doesn't engage, i.e. it's possible for LLQ traffic to exceed its bandwidth allocation in such a situation.)
Introduction: The "external-out enable" command is available for
configuration under the "router ospf process" in case of the IOS-XR
operating system. This command basically enables advertisement of
intra-area routes on the device as external routes in th...
IntroductionIn this article we'll discuss how to troubleshoot packet
loss in the asr9000 and specifically understanding the NP drop counters,
what they mean and what you can do to mitigate them. This document will
be an ongoing effort to improve troublesh...