Excess information rate
Excess information rate (EIR) is a parameter that relates to a virtual circuit (VC) in a frame relay service. It is closely related to the committed information rate (CIR). Where the CIR establishes a maximum throughput the service commits to carrying for a particular VC, the EIR establishes a maximum burst tolerance for that VC.
Operation of EIR
The EIR for a particular VC is the Be parameter (expressed in bits) divided by the Tc parameter (expressed in seconds). Because the default value of Tc is typically one second, EIR is typically Be bits per second. When a subscriber exceeds the CIR for a VC in a given Tc interval (typically one second), they are permitted to submit to burst to the sum of the CIR plus the EIR in that second. These additional frames will be marked discard eligible. If the subscriber bursts above the CIR + EIR in a given Tc interval, the additional traffic is typically discarded at the ingress switch.
Design Issues Related to EIR
Typically, the EIR is set to the difference between the user-to-network interface (UNI) access rate and the CIR. By doing this, the carrier is ensuring that there is no ability to burst above the EIR and the ingress switch will not discard frames at the ingress. This approach makes sense because most carrier core networks have adequate capacity for the traffic they support. When congestion occurs, it typically occurs at the egress because the customer has multiple VCs arriving at a single point.
Take the example of a classic hub and spoke deployment. If 40 branch offices have permanent virtual circuits (PVC) with CIRs of 32 kbps, and all of these connect back to a central location with a T-1 UNI, there is adequate capacity for the total of the CIRs (e.g., 1.28 Mbps). But if the branch offices all burst to full access rate, the total submitted traffic will be 2.56 Mbps. That is more than the T-1 UNI is capable of carrying. The egress switch will become congested. If the situation persists, frame discard will become necessary.
Note that this is a good example of a place where CIR + EIR might be less than the access rate. At the T-1 hub location, each PVC has a CIR of 32 kbps and an EIR of 32 kbps, for a total of 64 kbps. This is only 1/24th of the access rate at that location, leaving a lot of room above the CIR=EIR to burst. In this context, it is critical for the frame relay access device (FRAD), which is typically a router, to be configured to do traffic shaping. It should meter traffic onto the PVC to stay under the CIR + EIR total, because any traffic above that rate will instantly be discarded at the ingress switch. Better for the router to begin to backlog a bit, with the hope of being able to clear the buffers over time. If the router eventually congests, it can make the decision about what to discard locally.
|<mp3>http://podcast.hill-vt.com/podsnacks/2007q3/eir.mp3%7Cdownload</mp3> | Excess Information Rate (EIR)|