Aima A Derh

CLOUD PRACTICE QUESTIONS

COMPARISON
1. Cloud performance vs. dedicated servers

Dedicated always wins when it comes to performance

2. Cloud delivery models description and comparison
IaaS

Paas

SaaS

Descriptio
n

Includes raw IT
resources; must be
maintained/manage
d

Built on top of
preconfigured
resources

Provided as a
product
(software)

Control

High-level

No administrative
burden

Little to none

Usage

Full

Limited

Usage-only
(application)

3. Deployment models description, benefits, and comparison

Public

Community

Private

Hybrid

Definitio
n

Open to the
general
populous

Similar to
public but
limited to a
specific area

Owned/consu
med by a
signal person

Mixture of two
or more
deployment
models

Benefits

Available,

Shared costs,

Regular

Increased
1 | Page

Aima A Derh

accessible,
scalable,
cost efficient

multitenancy,
support and
maintenance
shared

benefits
oriented
towards the
consumer with
centralized
access

flexibility, great
for
transitioning,
fault tolerance,
control over
sensitive data

4. Advantage of cluster computing vs. grid computing: objective of each.

o Cluster: similar hardware, pools resources higher availability.
o Grid: dissimilar machine, complicated tasks large scale (can handle
more)
5. Disadvantages and advantages of cloud
Cloud
Advantages

Disadvantages

Cost: more resources for less money

Control & management: little to none

Availability/reliability: can access from

anywhere at any time

Trust: have to trust the provider

Disaster recovery: easily backed up

Security: everything can be taken

6. Disadvantages and advantages of a dedicated server

Dedicated
Advantages

Disadvantages

Control & management: on premise

Cost: maintenance/more resource means

$$$

Security

Accessibility: cannot be accessed from

anywhere

Performance: solely made for organization

Not easily expandable

7. Hardware independence, server consolidation, resource replication

definitions and examples

Definiti
on

Hardware
independence

Server
consolidation

Resource
replication

Puts both the OS and

installed softwares
configuration into a
single platform

Installation of multiple
OSes on a single
physical server

Standard file
operations: copying,
deleting, moving etc.

2 | Page

Aima A Derh

Exampl
e
8. Full vs. Para vs. OS-assisted vs. Hardware-assisted virtualization
Full

Para

OS-assisted

Hardwareassisted

How

VMM emulates
underlying
systems needed
by VM

Para-API used to
communicate
with VMM

Virtualizes at OS
level no VMM

CPU
virtualization
mode, sends
instructions to
VMM

Gue
st

Not aware

Aware

Aware

Aware

Info

High
performance;
Type-1/type-2 provide

Improved
performance

Benefits: low
VMM overhead
server density
Challenges:
cannot support
different kind of
guest

Benefits: better
performance
Challenges:
specialized
hardware (CPU)

9. CPU virtualization methods explanation and comparison

Full

Hardware

Para

Technique

Binary translation

Exit to root mode

Hypercalls

Guest
Modification

Unmodified

Unmodified

Modified for
hypercalls

Performance

Good

Good

Better

Used In

VMware, Microsoft

VMware, Microsoft,
Xen

Xen, some VMware

10.Characteristics and comparison of major commercial hypervisors

Characteri
stics

Hyper-V

KVM

Xen

VMware
vSphere

Can run full

version of
windows server

Converts Linux
kernel to
hypervisor

Can run
Windows and
Linux guests

Handles
scheduling/isola
tion of hardware
3 | Page

Aima A Derh

on each node
Compariso
n

Full
virtualization
Live
migration/high
availability

resources
Hardware with
virtualization
support
Para for guest
systems I/O
operations

Full/para
virtualization

Full
virtualization

11.VM storage: Thick vs. Thin

Thick

Thin

Definiti
on

Preallocated disk size

Disk size equal to amount of data

Pros

Greater performance

Efficient use of storage, easy to

clone

Cons

Wastage of storage, hard to

move/clone

Performance hit monitor over

provisioning

12.Traditional vs. software defined: differences, advantages, and disadvantages

Traditional

SDN

Difference
s

No central command and

control (CCC)

Network device DO implement

route controlling functions

Advantage
s

Resiliency

Ease of maintenance

Disadvant
ages

Scalability, security

13.SDN vs. network virtualization

SDN is NOT the same as Network virtualization!

SDN can exist without virtualization, and network virtualization can exist
without SDN

Network virtualization predates SDN

It is easier to virtualize an SDN switch

4 | Page

Aima A Derh
o

Run separate controller per virtual network

Partition the space of the flows

Open interface to hardware

14.Northbound vs. southbound API

Northbound API

Southbound API

Applications program network controller

Protocol between the controller and

device

No current standard

Current standard - openflow

15.Service-Oriented architecture; traditional vs. SOA architecture

16.Web service advantages

Web Service: a software system designed to support interoperable machineto-machine interaction over a network
Advantages:
o interoperability between various software applications
o leverage open standards and protocols
o

work through many common firewall security measures

17.REST; REST vs. WS-* (SOAP)

5 | Page

Aima A Derh

WS-* (SOAP-Based)
o Extensible; allows a variety of tools, features and protocols
o Allows automated transactions
o Interoperability problem
REST
o Lightweight; URI/URL is used as service locator
o Explicit control over resources
o Most implementations are limited to HTTP GET and POST
o Sophisticated designs will take longer

18.Strict consistency vs. eventual consistency: definition, comparison, and

necessity

Strict consistency is impossible to achieve in a distributed environment

Other consistency models exist - none scale well
Eventual Consistency maintains consistency for a single client respect to client's data stored
Replicas may temporarily contain inconsistent data
Updates are rare, all replicas will become consistent as updates are
propagated to other copies

19.ACID vs. BASE requirements

ACID
Atomicity: Transactions never see other
transactions
Consistency: must be in a consistent
state if database was in consistent
state
Isolation: Two transactions running at
the same time will not interfere with
one another
Durability: A committed transaction will
not be forgotten

BASE
Ignore atomicity: break a transaction
into smaller pieces and run them
parallel
Rely on soft state (cached data)

Send updates asynchronously

If parts of a transaction did not

finish, leave it and continue

QUESTIONS
1. What is a cloud service?
An IT resource made available over the cloud.
2. What is vertical and horizontal scaling?
1. Horizontal: scale in or out scale out by adding more servers
to the cluster
2. Vertical: scale up or down adding more memory or additional
CPU building up!

6 | Page

Aima A Derh
3. What
o
o
o
o

are the objectives of virtualization?

Isolation: what happens in one does not happen in another
Encapsulation: takes snapshot of the VM and recreates is
Portability: move VM from one place to another
Interposition: everything goes through the hypervisor
controlled

4. Is the expansion of cloud roles and boundaries good or bad, why?

Expansion of a trust boundary is bad. This is because you do not
want sensitive information to leak.
5. Inband or outband?
Sending within or outside.
6. What are hypercalls?
Direct calls to the hypervisor.
7. Why is scalability a challenge in SDN? Why is resilience a challenge in SDN?
Scalability is a challenge, if you move from traditional to new
environment more work needed, and can cause a single point of
failure (resilience).
8. How do you set up a backup mechanism in openflow?
1. Obtain affected flows
2. Find another path for each flow
3. Set up alternate routes
9. How does a flow table work?

o Works like an access list

o

Each flow table makes its own rules to decide what actions to
take with the packets (add, drop, etc)

10.What is Rest?
Rest: it is a multi-step access to what you want on the web
11.What is rapid local responsiveness?
Rapid local responsiveness is done by caching it is the verification
of inconsistent data. In order to make data consistent, this takes
place. It is not compatible for a large scale cloud as there would be
more nodes, it would be very slow.
12.What is Hadoop? Provide an example.
Hadoop: open source software for storing data and running

7 | Page

Aima A Derh
applications on clusters of commodity hardware (Ex. Graph mining,
machine learning, data mining, log analytics)
13.What is Hadoop stack? What is the role of this component?

Hadoop common

Contains libraries and other modules

HDFS

Hadoop distributed file system

Hadoop yarn

Yet another resource negotiator

Hadoop MapReduce

A programming model for large scale data

processing

14.What is Mapreduce? Provide an example.

MapReduce is a programming model for efficient distributed
computing.
You are given a basket of fruits. You separate the basket, and once
again, into multiple sets fitting the amount of people you have
(three people, three sets). Each person then calculates how many
fruits each person has and of what kind. Then you take that
information and organize per fruits. Finally you come to the
conclusion of how many fruits and what kinds there are.
15.What is Pig Latin? Provide an example.
High level language for map reduce (Ex. What assembly is in front of
C++)

8 | Page