[-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-]
Q1：
Consider the following relational design. Note that we give some functional
dependencies explicitly, but the design also obeys all dependencies that logically follow from the ones we state.
Consider the relational design with 4 relations given as follows:
R1(A,B,C,D) with dependencies A → B, C and A → D;
R2(B,E,F) with dependencies B → F and E → F;
R3(G,H,I) with no dependencies;
R4(F,G,J) with dependencies F → G, also G → J, and G → F.
For each relation, indicate all possible candidate keys, choose one primary key, and say whether or not the relation is in BCNF. You must explain your answer; also, for any relation that is not in BCNF, give a decomposition that is in BCNF.
[-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-]
Q2:
Consider the relational schema
author ( authorname, citizenship, birthyear)
book(isbn, title, authorname)
topic(isbn, subject)
branch(libname, city)
instock(isbn, libname, quantity)
Q2a) Give a relational algebra expression to compute the answer to the following
instruction: Give the birthyear of every Australian author who has written a book which is in stock in a library in Sydney.
Q2b) Write SQL to produce the views and permissions so that user Andrews can
see information about the books’ isbn, title, authorname and authors’ citizenship and birthyear, but only for books held in ‘Fisher’ library (and they can’t see information about books that are not held in that library). This user can also change the quantity of such books that are instock in Fisher library, but they can’t change any other information in the database.
[-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-]
Q3：
This question refers to a database containing (among others) two relations
Rel1(A,B,C) and Rel2(B,D). Each A field occupies 4 bytes, each B field occupies 8 bytes, each C field occupies 8 bytes, and each D field occupies 12 bytes. Rel1 contains 20,000 records and Rel2 contains 10,000 records. The files structures are as follows: Rel1 has a primary B+-tree index on A, with a dense B+-tree secondary index on B. Rel2 is an unstructured heap with no index. Assume that a block contains 1024 bytes, of which 24 bytes are used for header information.
Q3a) Calculate the space required for each relation (including the indices).
Q3b) Calculate the number of blocks of I/O needed to find the records of Rel1 with
a given value for the attribute B.
[-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-]
Q4:
Suppose our database contains three data items A, B and C each containing a
real number. Initially A=0, B=4, C=6. There are two transactions. T1 is Read(A); Read(B); Read(C); C := A+B+2; Write(C). T2 is Read(A); Read(B); temp := A; A := B; B := temp; Write(A); Write(B). The data is supposed to satisfy the undeclared integrity constraint C≥A+B. Note that each transaction, when run by itself, takes the database from a state where this constraint holds, to another state where it still holds.
Q4a) Suppose no locking is done. Give an interleaved schedule that violates the integrity constraint, and state what is the final value of the data items.
Q4b) Draw the serialization graph for the schedule you gave in Q4a) above.
Q4c) Suppose that each transaction uses strict two-phase locking. Place lock and unlock operations in the correct places in the schedule you gave for 45a) above. Show the first operation where the schedule will be blocked, when a transaction must wait for a lock it requests, and say which transaction holds the lock at that time.
------解决方案--------------------
R1(A,B,C,D) with dependencies A → B, C and A → D;