Sie sind auf Seite 1von 2

Database Management Systems

Exam 2 Review Problems


Question 1 Relational Operators
a)3 b)1 c)1 d)5 e)1 f)10 g)3,6 h)5,6
Question 2 Relational Query Optimization
1. Number of pages and tuples in each relation. Number of distinct values in Supply.sid,
Supply.pid, Supplies.city and Part.price. High and low values in Part.price.
2. (SU)P, (US)P, (PU)S, (UP)S
3. On S.sid, U.sid, U.pid, P.pid, S.city, P.price (most likely only a clustered one would be
used ).
4. Maintain plans that produce tuples in S.sname, P.pname order even if they are higher
cost. Final plan would need to include a sort (or hash) to perform projection.
5. As in (4) plans that produce tuples in S.sname, P.pname order even if they are higher
cost. Final plan would need to include a sort (or hash) to perform aggregation.
6. Number of distinct values of P.city (in addition to above).
All above join orders and (PS)U and (SP)U.
Also index on P.city.
Question 3 1)T 2)F 3)T 4)F 5)F 6)T 7)F 8)F 9)T 10)T 11)T 12)F 13)F 14)T
Question 4 1. Tables
create table Company (
companyId integer,
name char(50),
address char(50),
sector char(20),
primary key(companyId),
)
create table Price (
companyId integer,
date date,
value integer,
primary key(companyId, date)
)
2. Modi cations
(a) in the company table: sector not null
(b) in the company table: unique(name,sector)
(c) in the price table: foreign key(companyId) references Company
(d) in the price table: check (value =< 1000)
3. Views
create view transp(companyId, date,value) as (
select c.companyId, date, value
from price p, company c
where p.companyId=c.companyId and sector = 'transportation'
)
create view high(companyId,date,value) as (
select c.companyId, date, value
from price p, company c
where p.companyId=c.companyId and sector = ' nance'
and value > ( select avg(value)
from price p1, company c1
where p1.companyId=c1.companyId and sector = ' nance'
and p1.date = p.date )
)
create view sector(sector,date,price) as (
select sector, date, avg(value)
from price p, company c
where p.companyId=c.companyId
group by sector, date
)

Question 5 A)T B)F C)F D)T E)F


Question 6 1. Q1. There are many more tuples per page in Writer than in Paper. Hence, for Q1,
having a clustered index will make the query about 80 times faster than an unclustered. For
Q2, a clustered index is only about twice as fast as an unclustered index.
2. Let h be the height of the B+-tree index. Writer contains 50 pages. So a le scan costs 50
IOs. For an unclusterd index the cost is h + N. If N  50 - h the scan is better, otherwise
the index is better.b

Das könnte Ihnen auch gefallen