Give an example of a program at site A that requires the join of two tables, one at site B and one at site C. State the assumptions needed to justify the result that, as far as communication costs are concerned, the best implementation is to ship the table at site B to site C, do the join there, and then send the result to site A.
What will be an ideal response?
The three alternatives are
1. Send the table at site B to site C, do the join there, and then send the result to
site A. The total communication cost is the number of bytes in the table in site
B, plus the number of bytes in the join. (bytesB + bytesJ )
2. Send the table at site C to site B, do the join there, and then send the result to
site A. The total communication cost is the number of bytes in the table in site
C, plus the number of bytes in the join. (bytesC + bytesJ )
3. Send both tables to site A and do the join there. The total communication cost is
the number of bytes in the table in site B, plus the number of bytes in the table
in site C. (bytesB + bytesC )
The first alternative is better than the second if
bytesB + bytesJ ? bytesC + bytesJ
and better than the third if
bytesB + bytesJ ? bytesB + bytesC
Simplifying the math
bytesB ? bytesC
and
bytesJ ? bytesC
You might also like to view...
Where does a sparkline display?
A) In a single cell B) Always to the right of the data C) In a range below the data D) On a separate worksheet from the data
Given int num; the value of num is directly stored in its memory space.
Answer the following statement true (T) or false (F)