Ok, after trying to research the burst function of rate-limit on the web, I finally asked someone smart and found that my previous answer was grossly inaccurate and for that I apologize.

So here's the short answer - you will have the sum of the two rate-limit statements for total bandwidth. You could conceivably use the entire 2Mbps bandwidth + more (see below) sending from the two hosts specified by ACL102 PLUS 10Mbps bandwidth + more (see below) being sent from other sources permitted by ACL101.

And then here's the long answer. For the first rate-limit statement, you can send 375000 bytes in a 1.5 second timselot - You can determine the timeslot by taking ((Bc)*8)/avg-rate - in this case, (375000*8)/2000000. Where it gets complicated is calculating the effects of idle time. If the previous 1.5 second timeslot was completely idle, you can use all of those tokens in the current timeslot which would still give you an average of 2Mbps over 3 seconds. But you also have excess burst configured at twice the normal burst. Excess burst uses the previous two timeslots, so if both previous timeslots (the previous 3 seconds) were idle, you could send an additional 750000 bytes in the current timeslot. So, over the current 1.5 seconds (the first 3 seconds being idle), you could send up to 1,500,000 bytes(12Mb). Over the entire 4.5 seconds, you average 2.67Mbps. This means that those two hosts can send a maximum average output of 2.67Mbps.

Using the same calculations, the second rate-limit statement would afford you a max average output of 12.72Mbps.

This means that your maximum rate-limited interface output would be 15.39Mbps.

Does that help?