HW#3 --- last modified January 27 2019 04:57:45..

Solution set.

Due date: Apr 4

Files to be submitted:
  Hw3.zip

Purpose: To gain experience with query cost evaluation and query plan evaluation. To learn about database recovery techniques.

Related Course Outcomes:

The main course outcomes covered by this assignment are:

CLO4 -- Tune queries and know how to perform query performance evaluations

CLO5 -- Know database recovery techniques

Specification:

This homework will consist of two parts, a written part and a coding/experiment part. Both parts will be submitted in the Hw3.zip file. The written part should be in a file Hw3.pdf. For the written part of the homework you should write solutions for the following questions. In what you turn in, make sure to write the names and student ids for each group member. For each problem, first copy and paste the question, then write your solution to it beneath it.

1. For each of the following operations, write an iterator that uses an algorithm described in class to enumerate the output of the following operations: (a) distinct (b) bag difference. Make sure your iterator reads whole blocks at a time, but outputs tuples at time.
2. If `B (R) = B (S) = 25,000` and `M = 2000`, what are the disk I/O requirements of: (a) two-pass set intersection from class, (b) sort-join from class.
3. Come up with additional query parsing rules to add to our rules from the Mar 5 Lecture to handle SQL join clauses.
4. Consider the table:

   Y(c,d)	Z(d,e)
   T(Y)=600	T(Z)=400
   V(Y,c)=25	V(Z,d) =80
   V(Y,d)=60	V(Z,e) =100

   Estimate the sizes of relations that are the results from the following queries: (a) `sigma_{c=50}(Y)`, (b) $\sigma_{c=50}(Y) \bowtie Z$.
5. Assume `A=10, B=20` (here we imagine `A` and `B` are blocks that can hold 1 integer) are stored in a DB. Suppose a transaction does the following sequence of operations I(A), I(B), R(A,a), R(B,b), a:= a+b, b:=b+ 2*b, W(A,a), W(B,b), O(A), O(B). Show the undo log records needed for this transaction.

For the coding/experiment part of the homework, I want you to see the actually plan a DBMS will choose for various kinds of queries. You can conduct your experiments using the DBMS of your choice from among sqlite, Mysql, Postgres, DB2, and Oracle. So that the grader has something to evaluate I want you to produce a file Experiment.pdf which at its top a write up of your results, followed by transcripts of each of the explain operation I am asking for as well as how you created the input tables. Each of the databases mentioned has a variant on the EXPLAIN command which tells you how the database would evaluate a query. For example,

EXPLAIN SELECT * FROM USERS;

or some variant will tell you how the database would perform the query select * from users; without actually performing the query. To do the experiments you will first need to create a program DataGenerator.java. This program will be run from the command line with a line with the following format:

java DataGenerator start_value num wrap_number txtfile

The program should output into txtfile num many rows of two columns, space separated. The first column starts with the value start_value and increments one with each row. The second column's value is chosen randomly from between 1 and wrap_number For example, filling in these values we might write:

java DataGenerator 10 20 5 data.txt

The program might output into data.txt the rows:

10 1
11 1
12 5
13 2
14 3
15 3
16 4
17 5
18 1
19 2
20 1
21 2
22 4
23 1
24 5
25 2
26 3
27 4
28 4
29 1

You should include both the file Experiment.pdf and DataGenerator.java in the Hw3.zip file you submit. Create five tables R1(A,B), R2(C,B), R3(D,B), R4(E,B), R5(F,B). All of the columns be of integer type. You should include the execution of the create tables in Experiment.pdf. Next generate five tables worth of data with the following lines:

java DataGenerator 0 500 50 R1data.txt
java DataGenerator 500 500 50 R2data.txt
java DataGenerator 1000 500 50 R3data.txt
java DataGenerator 1500 500 100 R4data.txt
java DataGenerator 2000 500 100 R5data.txt

Use the bulk loader facility of the database you chose to load these five files into their corresponding table. Copy the text or show a screenshot of performing the load operation into Experiment.pdf. Now consider the join of all five tables with the condition R1.B=R2.B and R2.B=R3.B and R3.B=R4.B and R4.B=R5.B and R5.B=51. Express this query in SQL in as a join that would look like a left-tree, bushy tree, and right-tree. For each way, use explain to find out how your DBMS would execute the query. Put this information into Experiment.pdf as well. Try executing each of your equivalent queries and check that the results match. Write up, again in Experiment.pdf, which you think is the best way to do the query. If you desire you can also experiment with indexes to see if it helps the execution speed.

Point Breakdown