Design a query plan for the join example in Section 16.3.1, assuming the same table sizes as in that section, but with the following differences:

a. An application at site B requested the join.
b. An application at site C requested the join.

a. 1. If we send Student to site C, compute the join there, and then send the result to site B, we have to send 540,000 bytes (= 15,000 ? 12 + 20,000 ? 18).
2. If we send Transcript to site B and compute the join there, we have to send
300,000 bytes (= 20,000 ? 15).
Thus alternative 2 is better. If we also allow the alternative of the semi-join described in that section, except that after the join is performed in step 3 of that protocol, the result is sent to site B, we have to send 465,000 bytes (= 45,000 + 60,000 + 360,000), as described in that section Thus alternative 2 is still the best.

b. 1. If we send Student to site C and compute the join there, we have to send
180,000 bytes (= 15,000 ? 12).
2. If we send Transcript to site B, compute the join there, and then send the
result to site C, we have to send 660,000 bytes (= 20,000 ? 15 + 20,000 ? 18).
Thus alternative 1 is better. If we also allow the alternative of the semi-join described in that section, except that after the join is performed in step 3 of that protocol, the result is just kept at site C, we have to send just 105,000 bytes (= 45,000 + 60,000).
The semi-join is thus the best alternative.