Distributedsys Updated
Distributedsys Updated
Distributedsys Updated
l
?'1rj 1"'Resu&r"''' r'''
TRiBHTIVAN UNIVERSITY Exarn. ":-
'"'': ':: '
Df-
INSTITUTE OF ENGI}'{EERING .{,evel ,L}L Full Marks 80
) Discuss the fiinctionalities provided by RMI software. How is the event and notification
system igplemented in distributed object based communication? [3+5]
3. Whai is distributed file system? Explain the principle operations of any one modern
distributed file systemt {2+7}
4. What is ihe issue in Lamport's tirnestamp? How do you avoid the issue? Explaio with
your altemate algorithm. [2+8]
5. How does a ftew coordinator elect in exeouting central coordinator algorithm? How to
come to consensus in DS? Explain. [5+5]
6. What is fault? llow to implement primary-backup repiioa system? How is it differ from
active replication? l2+4+2J
7. What do you mean by forward and backward recovery? How to implement coordinated
check pointing for recovery in DS? t7+61
1. What are the major goals of distributed system and what are the challenges during the
design of distributed system,
2. Define clistributed objects and explain communication between distributed system.
3. Define Distributed file system. Point out the differences between stateless and stateful
services.
4. Explain components in CORBA architecture with a diagram.
5. Explain Lamport's logical clock with its pros and cons.
6ri. Explain reliable multicast with its properties and an algorithm.
J. Specify Data Centric consistency models and explain any one of them in detail.
8. Why is it necessary to maintain transaction? What is a deadlock and what are phantom
deadlocks.
.9. What is fault tolerance? Explain different type of faults that may occlu in a distributed
system.
10. Write short notes on any two:
a) RPC
b) Monolithic and micro kernel
c) Mach
,lr*
TRIBI-IUVAN UNiVERSITY Exam. .:''t' .i":', ' Rogrrlilt'/:llacirr ' ...': , .
Candidates are required to give their answers in their own words as far as practicable.
Attempt AI! questions.
The figures in the margin indicate Fqll.,Maflp.
Assume suitable data if necessary,
1, Why distribtrted system is preferred over centralized system? Explairi Tfid layers of
transparency. l4+4)
) What do you mean by RMI software? Comparatively discuss RPC with RIIII. [3+5]
a
Compare Stateful and stateless service, Describe the architecture and operation of
SLINNFS with its services. 12+61
7. How replication is used as a basic scaling technique in distributed system? Explain the
active replication model with its advantages and disadvantages. 12+61
8, Compare nested transactions and distributed hansactions. Explain the two-phase commit
protocol of handling distributed transactions. [2+6]
9. What do you leam from Byzantine generals problem? Explain the basic principle of K-
fault tolerant. [3+5]
10. Write short notes on: (Any two) [4+4]
a) Reliable Group Communication
b) Distributed deadlock
' c) Forward and Backward relovery in distributed system
{.**
it tt
\t
\t
lrt
I
33 'TRIBHU\TAN LINIVERSITY
2. Why naming is necessary in distributed system? Explain Sun Network File System
architecture with its features. 12+61
3. What is DNS? Explain the DNS working mechanisms with suitable example. [2+61
4. What do you mean by DOS (Distributed Operating System)? Briefly explain the
system.
Monolithic and microkernel architectures of operation I2+4J
6. What is Network Time Protocol (NTP)? How Berkeley minimizes the problems of single
time server failwes of Chistian's algorithm. [2+41
7. What is the need of an election algorithm? Explain non token based Ricart-Agrawala
mutual exclusion algorithm along with an example. 12+61
8. Differentiate between passive and active replication approach. Discuss with a technique
that make the distributed system hig?tly available. [3+5]
g. Write down the rule of two-version locking. Explain how Optimistic concurrency control
mechanism works? L2+61
10. How does triple modular redundancy works? Explain how reliable client seryer
communication can be achieved in distributed system. [4+4]
11. Write short notes on: (any two) [3+3]
i) . Lamport's clock
ii) TlB/Rendezvous
iii) Feedback suppression mechanism in M'cast communication
{.**
,1
Full Marks
Pass Marks 32
2074 Ashwin tffi'narr* iiviT 3 hrs.
f"rargsl.:pi$bgtq{.i# *
candidates are required to give their answers
in their own words as far as practicable.
Attempt All questions.
TheJigures in the margln indicate
{ Assume suitable data if necessary.
fuil MsrE.
8' [3+5]
How cascading aborts occurs and solved? Explain
three phase commit protocol with state
diagram.
9. what is K-fault tolerant system? Explain fault recovery t8l
techniques [2+6]
10. Write short notes on: (Any two)
:f :F*
35 TRIBHUVANUNIVERSITY
INSTITUTE OF ENGINEERING
Examination Control Division
2073 Shrawan
{ Candidates are required to give their answers in their own words as far as practicable.
Anempt All questions.
Thefigures in the margin indicate FuIl Marks.
Assume suitable data if necessary.
l. What are the principle applications of Disributed System (DS)? Discuss the advantages
and disadvantages of DS. [3+s]
What are the requirements of Distibuted File System? Describe file service architecture
for Disfiibuted File System. [4+4]
3. a) Why naming is necessary in distributed system? Explain Domain Naming Service
(DNS) with its features. [1+4J
b) What are the advantages of micro-kemel over monolifhic-kernel? In your view, which
kernel is preferable for distributed Operating system and why? 12+31
PO
PI
P2
b: i;
Define distributed coordination in DS? Explain how token ring algorithm works for
mutual exclusion in DS. 12+6]
7. Define replication and fault tolerance in DS and explain why are they necessary? Explain
how replicatfon enhanced scalability for DS. 16+21
What is LOCK and DEADLOCK in DS? Discuss the methods of distibuted deadlock
avoidance. [4+6]
9. Write short notes on: [2x31
i) Process Resilience
ii) Mach
TRIBHUVAN TINIVERSITY
INSTITUTE OF ENGTI\{EERING
Examination Control Division
2072 Chaitra
Candidates are required to give their answers in their own words as far as practicable.
Attempt AII questions.
Thefigures in the margin indicate Full Marks.
Assume suitable data if necessary.
1. Differentiate between centralized and distributed system? Explain the design issues
related to distributed system. [2+61
2. Discuss the importance of Distributed File System (DFS). Describe the operations of
SLTNNFS with its properties. [2+61
a
J. Explain RMI with suitable diagram. How RMI is superior to RPC? [8+2]
4. What is the role of middleware in DS? Explain about CORBA and its services. [2+8]
5. Differentiate between physical clock and logical clock. Why it is diffrcult to synchronize
physical clock? Describe a method for physical clock synchronization. 12+2+67
6. What are the basic requirements for mutual exclusion in distributed system? Explain the
non-token based distributed mutual exclusion algorithm and compare it with token based
algorithm. [2+81
7. What are the reasons for Replication? Explain active replication model with its
advantages and disadvantages. [3+5]
8. What do you mean by nested transactions? Explain optimistic concurrency control
method with its advantaqes over other concurrency control methods. [2+4+21
9. Write short notes on: [4+41
i) Distributed OS
ii) JiNr
in distributed transabtions. [4t-Gj
35 TRIBHUVANUNryERSITY Eram. I{cgu l:t r'
INSTITUTE OF ENGINEERING LeVel BE Full Marks 80
I
35 TRtsHWAN UNIVERSITY Exam. Nerv llack & I-ttcr
. INS-TITUTE OF.ENGINEERTNG_ Level ----- BE EuILMerks- J0 ",
1. How do you define DistibutedSystem? Explain with the model, how hardware, data and
controls are distributed in the distibuted system environment. .,i 12+61
2. Explain the ways how distibuted objects communicate with each other. Differentiate
betrveenRPC and RMI. ' 14+47
3. Define distibuted file system. Draw and explain distributed file service architecture.{ ,... ,
detail. i;g112+61
i,i'r j'
':'
P.2
. .al
35 TRIBHWA}'TTJNIVERSITY Exam.
ENGINEERINq
INSTIIUTE OF , f,evel +E
- -Examination Controt Division Programme
2069 Chaitra Year / Part rV /I
- Distributed
/ Candidates are required to give their answers in their own words as far as
{ Attempt All questions.
{ The figures in the' margtn.indicate Full Marks.
{ Assume suttable data if necesiwy.
t
1. What are the major'.goals of distibuted system? Explain the need-of transparency tB' - -'-:-'
distibuted system along with the challenges in achievingthat. 14+41 .
2. How do you convince that middleware plays the important role in Dishibuted System?
Explain the operation of RPC in client seryer comnnrnication in Distibuted System. [3t5]
3. What do you meanby file and direetoryservice? Explainthe operation of SUNNFS with
[3+s]
4. Why network operating system (NOS) is widely preferred over disuibuted operating
'- ' [4+4]
system (DOS) in practical distibuted systems? Explain DOS as a middleware.
5. Define logical and physical clocks. Explain Lamport timestarnp algorithm along with an
exarnple. 12+6)
6. Present a practical scenario where you need an electiod algorithm. Explain an efection
algodthm with example that is suitable to your scenario. \ [2+4]
7. Compare passive replication with active replication approach. Also discuss with a
technique that make the distributed system service highly available. 12+41
8. What do you mean by Distibuted Deadlock? Explain the two-phase commit protocol of
handling distributed transaction. ,
[2+5]
9. Wbat are the flat and nested tansactions? Describe the methods for concurrency contol
in distributedsystem. '
[3+4]
10. What do you mean by faults, failures and errois? How do you handle faults in Distributed
. brief.
System? Explain process resilience approach in 12+2+41
I i. What is IDL? Explain CORBA RMI with its services L2+41
'rfrIt
P.1
q,{
"8; Sfi'y do we hsve single rrr#sr in GFS managing millions of chunk seffers? What Ef,r: r
5. Srrypse your cosrmtly got more invesfinent so ycu can now incre*se ysur defax$t
reptication order by 2 fum current vdue 3 so as to mske it your data more avaitrable and
reliable. Suddenly master has to create two replicas of erch file chunks. Yet, it has to
satisfu client request wtrich are more important than just replication. How does single
master manage re-reptricafiontask? ',:
LtI
l
6: Lipt urd explain four main mastsr ope,ratioru in GFS: - [4]
ll
'?. Expl*in twnr TrloSQL'. Why does nonnolization ftiI in data analytics scenmio? [4+4I
E. Uefrre the oompoaents that ur*ke up a hasic MapReduoe job and illustraE with diagram how :
dsts ttre data flow though llailoop MapRoduco. ttl
1.. .i
v
':i . .u tj.-r
,. i; 'f I;.ilri
i it
I
,
\
- Distributed (Electtve I)
,/ Candidates are required to give their answers in their own words as far as practicable.
'/ Attempt fllquestions.
,/ All questions carry equal marks.
./ Assume suitable data if nec€ssar!:
8. a. Write pseudo code for inverted index (term vector per host) generation through
mapreduce
I ' Discuss the important characteristics ofDistributed Systems, Explain distribution tanspareny goal of
distributed systems (3+7)
2. . What is an architectural style? Disouss with one example structured P2P architecture. What is a
super peer, discuss its usage in P2P systenls? (2+4+4)
3. Discuss assumptions made in GFS design and their consequences. Explain the metadata of
GFS? (6+4)
4'. Explain data and control flow and working of lease mechanish in normal write operation of GFS.
(10)
5. ' Discuss the main points that MR drarvs from referential transpareny in functional programming.
With an example of your choice explain the MapReduce programming model. . (3+8)
6. You are given student records consisting of following fields in each record: (t0)
rollno(integer),department id (inJeger), name (string),subject code(integer),score on the subjeot (double). You
can assume the delimiter of the fields as space. It is requiredto calculate the agerage value of score for a
student and present the average score and total score along with name, department and rollno of the shident
in the output records. The order of the output be according to rollno, and if the same rollno appear in two
departments then by rollno+department. (Note here that the rollno and department id are integers not strings).
t
Wrlte a complete code for mapper,reducer,combiner(if any),partitione(if any)in Java language on
Hadoop platform. You can ommit theJob submission and imports part. Clearly speoiff the key/value data
tvpes [n each phase.
OR
Write a complete pseudocode for mapper,reducer,partitione(if any),combinu(if any) along with clear
explanation ofthe key/value types in each phase. (10)
'\i\-' 7. Discuss the problems associated with parameter passing in RFC. Explain with diagram different
RPC.
alternatives of asynchronous (6+4)
8, Explain any 3 of the following (3X3)
a) Use of multicast communication in Disnibuted systems
b) External sort is related with MR programming model in Hadoop
c) Compression can improve the performance of MapReduce job
d) P2P style of architecture of Cassandra (ust discuss to show the p2p style not the entire
architecture)
**:f