Have you every thought that you knew a topic pretty well but then someone uses terminology that you aren’t used to? People that use Cisco a lot or live outside the MEF world use another terminology than people that are working on MEF certified networks. Even if we both know the concepts, if we don’t speak a common language it will be difficult to communicate and to the the right end result.
When I took the CCDE written at Cisco Live, some of the QoS related material felt a bit off to me. I feel quite confident with QoS so this took me by surprise. My theory is that some of the material was written by someone coming from another background and uses some wording that just felt a bit off to me. I thought that I would read through some of the MEF material to broaden my QoS horizon and see what other terms are being used. At the very least I will have learned something new.
If we start with the basics, we have flows in our networks and these flows have different needs regarding delay, jitter and packet loss. I will write different terms and I will indicate which belong to MEF terminology, the other terms will be related to what Cisco calls them or what they would be called in general outside of the MEF world.
Delay
Latency
Round Trip Time (RTT)
Frame Delay (MEF)
These all relate to how much delay that is acceptable in the network. It may be one-way or two-way requirements depending on the nature of the traffic. RTT always refers to the two-way delay.
Jitter
Frame Delay Variation (MEF)
The MEF term is actually a bit clearer here as jitter is the variation of delay.
Packet Loss
Frame Loss Ratio (MEF)
Once again, MEF term is a bit clearer because we are interested to see the packet loss as a ratio, such as 1/100 packets which we then use as a percentage for what is acceptable loss on a circuit.
Commited Burst (Bc)
Commited Burst Size (CBS)(MEF)
The Bc or CBS value is used to define how much traffic in bits or bytes can be sent during each time interval. Picking a too low value can lead to customer dropping a lot of packets and picking a too high value can lead to long time intervals which could affect high priority traffic. The formula Tc = Bc / CIR can be used for calculations.
Burst Excess (Be)
Excess Burst Size (EBS)(MEF)
Be or EBS is normally used to provide the customer a more “fair” use of a circuit by allowing them to send unused credits from one or more previous time intervals. This means that they can burst momentarily until they have used up the Bc + Be credits.
Committed Information Rate (CIR)
This is the rate that is guaranteed to the customer in the contract. The physical line rate could be 100 Mbit/s but the CIR is 50 Mbit/s. It should be noted that this is an average rate and that traffic is always sent at line rate which produces bursts of traffic. This means that the customer will for short periods of time send at above the CIR rate but on average they get CIR rate on the circuit.
Excess Information Rate (EIR)(MEF)
A provider/carrier may allow a customer to send at above CIR rate but only those packets that are within the CIR are guaranteed the performance characteristics as defined in the SLA. This is commonly implemented with a single rate Three Color Marker (srTCM) where packets that are within the CIR/CBS are marked as green, packets above CIR but within EIR/EBS are marked as yellow and packets that exceed the EIR/CBS are marked as red. Green packets are guaranteed performance as defined in the SLA, yellow packets get delivered according to best effort and red packets are dropped.
This illustration shows the concept of srTCM:
Peak Information Rate (PIR)
As noted by Faisal in the comments. PIR is not the same as EIR. PIR is actually CIR + EIR which means that we have two token buckets filling at the same time and incoming packets are checked against both to see if it matches CIR rate or EIR rate which will then set the color of the packet to be green or yellow. One example could be where customer has CIR of 10 Mbit/s and EIR of 10 Mbit/s which gives a combined rate (PIR) of 20 Mbit/s. The first 10 Mbit/s is guaranteed and the other 10 Mbit/s is sent through the provider network as long as there is capacity available.
This is a short post on different QoS terminology. Which terminology are you most used to?
Good article and choice of subject!
Few Points as additional clarification !
1. CIR is the average rate guaranteed to a customer, this is erroneously taken as a constant rate. sometime. It means the customer rate can can fluctuate above and below CIR as long as the average rate is within CIR
2. There is a difference between PIR and EIR. PIR is an IETF term, EIR, a MEF term.
PIR = CIR + EIR
Example: CIR= 100 Mb, EIR= 10 MB, in that case PIR would be 110 MB.
I have seen quite a confusion with the people using PIR and EIR interchangeably , which is not the case.
Faisal
Thanks Faisal! Yes, people often don’t understand that traffic is always sent at line rate no matter what the shaper/policer is configured at. I’ll try to clarify this. Good catch on the PIR, I will correct that as well. Thanks!
Daniel, This is an excellent blog post! Since finishing my CCIE R/S, I really enjoy these type of side-by-side comparisons between Cisco and other vendors etc.
Thanks Mark!
From a design standpoint, the enterprise network manager must be made aware of the performance capabilities of the entire Cisco product line from the low end teleworker router to the campus crypto and WAN aggregation to deploy a device capable of processing the expected offered data load for the configured security, management, and control plan of each device. This section defines the key elements of the network topology, including terminology and definitions.