Ethernet Services and Committed Data Rates

user-pic

Having a Committed Data Rate (CDR) of some for is absolutely critical for all packet-based services. However, as the services become more complex, the parameters used to measure and enforce the CDR become increasingly complex. This is especially true for overhead-intensive protocols such as IP and Ethernet. Do you pay for data based on the entire packet size, for instance, or just the payload? Join us to explore these issues in this discussion.

2 Comments

| Leave a thread/comment
In the information about Global Crossing's Ethersphere service there is a mention of various Committed Data Rate (CDR) options. This brings up an interesting question in general. This question is exactly what is measured in the CDR? Is it the entire size of the Ethernet packet or something more efficient - such as the size of the MPLS packets transmitted across the network? The reason that this is important to know is that Ethernet has a massive (relatively speaking) amount of overhead, especially since the entire IP overhead must also be accounted for. So a CDR of 10 Mbps (for example) is vastly different with and without the Ethernet packet overhead. Historically, one of the great debates about WAN services (and CIR and equivalent measurements) was the efficiency of the protocol. For instance, it wasn't all that long ago that we were concerned (rightly or wrongly) about the 10% "cell tax" that ATM imposed on the network. Now we're transmitting entire IP packets over the network. And if we're transmitting entire Ethernet packets over the net, then we're transmitting the TCP/IP (or UDP/IP) AND the Ethernet packet over the net. I realize that bandwidth has become less expensive over the years, but are we really ready to send all of this over the network? The bottom line, imho, is that you need to make sure when you're communicating with your service provider about a CDR, you know what is and what is not included in that CDR.
For a basic understanding of what is and what is not included in a CDR/CIR, imho there is nothing that compares with FRF.13 and the fine work done on that document 11 years ago. It's still available at: http://broadband-forum.org/technical/download/FRF.13/frf13.pdf

Leave a thread/comment

Please note that commenting here will start a new thread unless you click on "Reply" next to a specific comment and the "Replying to" text appears.

Using HTML is highly recommended, as comment text is otherwise unformatted. Click here to use an automated HTML tag generator. Supported tags are listed below.