Just wanted to get opinions, if its advisable to apply QOS on routed gigabit interfaces on 3560s? We've got a couple of WAN locations (using free space optics) hooked up thru our gigabit SFP ports, and while I haven't seen any QOS issues yet; there's been questions as to if its necessary to apply QOS to those interfaces? Our Max traffic has, so far, been around 60 to 100mb/s but the connections are capable of going up to a 1gb full duplex
Second question, not quite related, is there a point where applying QOS doesn't make sense?
I would recommend applying QoS to these links even though you have more than enough bandwidth. This is especially true if you will have real time traffic such as voice/video traversing the link. In your scenario, the gigabit link can basically be treated as a LAN and QoS applied accordingly. Take a look at the "Campus QoS Design" section of the QoS SRND below. In my opinion, the only point at which QoS doesn't make sense is if you are only dealing with best effort traffic.
I realize that, what I'm asking for is there a certain point where the speed of the link negates the need for QOS? Where traffic on the interface goes thru so quickly, that you'll be making things worse by adding QOS.
For example, once your house is flooded, it doesn't matter how much water came in. Your house is still flooded.
My prior answer may seem obvious, since you "realize that", but perhaps you don't fully see all the significance within what I wrote when you ask whether the speed of the link negates QoS.
". . . what I'm asking for is there a certain point where the speed of the link negates the need for QOS?"
If by QOS, you have in mind the application of something beyond the often normal FIFO tail drop queue for an interface, and for fixed amounts of traffic bandwidth, such that the aggregate utilization is always less than the link's maximum bandwidth, the answer is yes.
Any combination of 80 Kbps CBR VoIP flows and FTP flows sourced from 100 Mbps connected hosts, all across a 1 Gbps link where the aggregate bandwidth doesn't exceed 1 Gbps, shouldn't need QoS.
If the aggregate bandwidth can exceed the transit link's bandwidth, you'll need to analyze the situation.
One 80 Kbps CBR VoIP flow and one FTP flow sourced from 1 Gbps connected host, both across a 1 Gbps link, where the aggregate bandwidth exceeds 1 Gbps, would need analysis.
What might also help, and perhaps where bandwidth does come into play, is the actual instantaneous delay when there's congestion.
Nine 1500 byte FTP packets in a queue, followed by a single VoIP packet, will make for more serialization delay on a T1 vs. 100 Mbps Ethernet (you can calculate the delay for both).
The extra delay on the 100 Mbps link might be within the acceptable bounds for the VoIP packet, but unacceptable for the T1. This is the kind of analysis needed when there's any congestion. The congestion can be transient, such that the link is only has 20% utilization, yet the issue is still the VoIP packet behind other packets during traffic bursts.
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...