APJ ABDUL KALAM TECHNOLOGICAL UNIVERSITY Previous Years Question Paper & Answer

Course : M.Tech

Semester : SEMESTER 1

Year : 2018

Term : DECEMBER

Scheme : 2015 Full Time

Course Code : 01 CS 6103

Page:2





PDF Text (Beta):

b. Assume that in two-phase locking, read locks are granted and released as and (4) when

needed, but write locks are granted in the growing phase and released in the shrinking
phase only. Will this new protocol ensure serializability? If yes, justify. If no,
give a counter example.

1
PART B

4 a. Suppose that in the two-phase commit protocol for distributed databases, phase 1 is (5) modified

டி

7.

as follows: when receiving the message, each participating site which is ready to
commit sends a message but does not write a record to the log. Will this
modified protocol work correctly always? Justify your answer.

b. Illustrate how semi-join leads to efficient query processing in distributed databases (5)

a. List out important data partitioning schemes in parallel databases. (2) 1.
Suppose that we have 4x4 matrix of processors labeled as Pll, P12, P13, P14, (8)
1121, P22, P23, P24, El, P32, P33, 1134, 241, P42, P43,
1344 where Pij represents the processor at row i, column
j. There are two relations R with 4000 records and 5 with
400 records. R is divided into four equal partitions and
stored in Pll, P21, P31, 241. Similarly, 3 is parationed
into four equal partitions and stored in Pll, P12, P13,
P14. Compute the communicaHon cost for asymmetric and
symmetric versions of fragment-and-replicate join.
Communication cost 15 the total number of records communicated between
processors as part of join. Assume that the relations are not re-partitioned as part of
the join process. Ignore the cost of result storage, transfer
etc. List out any assumptions made.

a. What form of parallelism (inter-query, interoperation
or intra-operation) is (3) likely to be most important for the following
tasks 010. * / ° ! प्र-प्र . kilonline.com

i) Increasing the throughput of a system with many small queries

ii) Increasing throughput of a system with a few large queries, when the number
of disks and processors is large

Explain how deadlocks are handled using local/ global wait-for graph in a distributed
(4) environment? What is false cycles?
Write a brief note on spatial data management (3)
PART C
a. VVhat is the difference between mutable and immutable objects? (3) b. Write INFORMIX

DDL statements to create a table with row objects to store details of (7) students which include
rollno, name, age, courses_taken. All except courses_taken are simple attributes. courses_taken is

Similar Question Papers