Most monitoring tools base "total bandwidth " on a T1 at 3088, double the speed. This I suspect is based on the premise that Frame-relay can do full duplex, but there are no guarantees that you can do true full duplex (1.544 in and out at the same time.)
Most monitoring tools have to assume some baseline as their reference. The specification for T1 at full duplex at Layer 2 (frame) is indeed 3088 kbps (1544 kbps in each direction.) So this is a natural assumption since they don't have access to any other measure of the link capacity. And that is their limitation.
But, as you have noted, just because it says T1 on the outside doesn't mean that it is functioning to specification on the inside. To know precisely what any particular network segment will support, you have actually measure the pipe. How to do that?
Well, suppose you stuffed a file down the link as quickly as you can and measure how long it takes. If you are competing with other traffic, or the FTP client isn't fully optimized, or your NIC driver is sub-optimal, you won't get the full capacity. So it is unlikely that you will be able to find out that way. However it is common practice - for example, for customers wanting to know their link capacity, many ISPs offer timed downloads from Web pages.
Are there ways to do it more reliably? Absolutely. One approach would be to use a precision flooding tool that can fill the pipe completely ? for example, a SmartBits generator at each end. This will stress the link to its capacity and thus you will know what it can reliably do in raw bits per second. However, this is expensive to do and time-consuming to set up. And you can't always flood the network just to measure it. And there is no way to measure utilization since this method will clobber any traffic on the link.
An alternative is to use a non-intrusive sampling method such as we (jaalaM Technologies) use in appareNet, our network performance analysis system. The software sends very precisely controlled bursts of packets down the link from any host it is installed on, detecting both the total capacity of the link (in both directions), as well as the presence of other traffic. The sampled measures are treated analytically to extract highly accurate measures of both maximum achievable bandwidth and the relative utilization. And, I might add, all this in a few minutes and without deploying any agents.
Done this way, utilization is precisely in terms of what the link can actually support. And
specific to what any application using the link would see. No guessing or relying on the
theoretical specifications. And that, I gather, is what means the most to you.
For more information on measuring bandwidth, read this tip series:
This was first published in March 2003