NET302 (1)
NET302 (1)
Plymouth University
STAGE FINAL
INSTRUCTIONS TO CANDIDATES:
Section A – Answer ALL of the questions. This section is worth 40 marks in total.
Section B – Answer TWO questions from the FOUR available. Each question is
worth 30 marks.
Candidates are not permitted to look at the examination paper until instructed
to do so.
Semester 1 Exam
Q1. Select and list three challenges that distributed systems must face. Explain each challenge
and indicate the extent to which it is a problem for systems that are not distributed.
(9 Marks)
Q2. On the face of it, a 3-tier system is more complex and involves more protocols than a 2-tier
client server system. What advantages does the 3-tier system provide?
(8 Marks)
(a). Describe the mechanism of an RPC. Your answer should involve a diagram and
should explain the terms marshal, client stub/proxy and server skeleton/proxy.
(8 Marks)
(b). Describe different ways in which a RPC can deal with repeat requests.
(3 Marks)
Q4.
This question concerns transactions.
(Over/. . . )
NET302-2017/2018 Page 2 of 6
Section B: Choose TWO questions to answer
Each question is worth 30 marks.
Q5.
(a). A system is described as 99.99% available; compute the expected downtime in one
year.
(4 Marks)
(b). Explain what replication means and how it increases the “mean time to failure” for a
distributed system.
(4 Marks)
(c). A certain small company regards its web presence as mission critical and therefore
arranges to have two of each of the following:
For each item in the list above, describe how the system could be made fault
tolerant. In your answer, indicate whether the method you describe would also
support load balancing.
(12 Marks)
(d). For load balancing between geographically distributed servers, DNS is often used.
This can be done with varying degrees of sophistication. Describe how the simplest
DNS load balancer operates and then explain what additional features more
sophisticated systems provide.
(10 Marks)
(Over/. . . )
NET302-2017/2018 Page 3 of 6
Q6.
(a). Describe the messaging communication paradigm and list the advantages and
disadvantages it holds over RPC-like communication.
(10 Marks)
(Over/. . . )
NET302-2017/2018 Page 4 of 6
Q7.
(a). Briefly explain each of the terms IaaS, PaaS and SaaS and in each case give an
example of such a service.
(9 Marks)
(b). Server colocation hosting has been available for many years but cloud computing is
a fairly new phenomenon. Describe the cloud computing service provided by
Amazon or Rackmount in such a way as to make a clear contrast with the service
provided through traditional colocation hosting.
(6 Marks)
(c). Draw diagrams to illustrate the architecture of the operating system, virtual machine
monitor/hypervisor and hardware when an application runs in a:
(Over/. . . )
NET302-2017/2018 Page 5 of 6
Q8.
(a). Churn is a problem for large internet scale peer-to-peer systems in which the peers
are separately controlled by end users. Give an example of churn in this context.
(4 Marks)
(i) Explain how reduced churn affects the way that messages are routed between
peers by contrasting with the way this is done in the situation in part (a).
(ii) Why is churn not completely absent in this context?
(8 Marks)
(c). Peer to peer systems have evolved through three generations of approaches to
search for content that could be called: centralised, flooding and distributed. Briefly
describe the first two approaches and explain the advantages of the most recent
generation, the distributed hash table (DHT) approach.
(10 Marks)
(d). The relational model for databases emphasises consistency above availability and it
is hard to horizontally scale such databases. Alternative “NoSQL” approaches to
structured data storage have been proposed. Explain the limitations imposed by the
CAP theorem on these alternatives and name an example of a system that does not
follow the relational model.
(8 Marks)
END OF EXAM
NET302-2017/2018 Page 6 of 6