CLOUD COMPUTING- CASE STUDIES

CLOUD COMPUTING- CASE STUDIES


the current lecture on cloud is on some case
studies and more specifically case studies related to the available cloud platforms and
including the cloud simulation platforms so we are going to go through some of them so
before i start i would like to mention that you know before even before the adoption of
cloud whether it is for r n d purpose or whether it is for actual use in a business it is required
to assess this particular technology the different solutions that are offered by this technology
it is required to assess so how can we assess one of the ways is through simulations so
there are different simulation platforms that have been that are made available by different
community groups so we can use those in order to assess how cloud is going to perform what
are the different modules what are their implications on the overall performance of the system and
so on so that is a very important aspect of cloud computing
so i will go through some of these different cloud simulation platforms but before that
i would like to ah make you understand some of these issues of simulation so first of
all the simulation tools would be required to ensure reliability scalability and repeatability
for of performance evaluation so you know we have to ensure that the system is reliable
even before it is used it is scalable and the data that we are getting you know that
is repeatable so we have a repeatable environment which will give you the same or similar kind
of data even if you rerun the same simulation over time you know you you know you run in
you know at different instants of time you know you have a repeatable environment which
will give you similar kinds of data so that is very much required for performance evaluation
the simulators basically facilitated pre deployment tests of services then that is basically quite
generic as well i mean not just specific to cloud any simulator even mean what a simulator
does is even before you deploy the infrastructure the services is the platform as a whole you
need to test how it is going to work so this is another purpose of the simulator and the
third is as the demand of cloud computing is going everyday the simulators and technologies
are needed to study how things how different features which are going to be enabled which
are going to be added additionally how they are going to perform in the cloud environment
so cloud simulators allow the customers to evaluate the services to test the services
and the platform at no additional cost because no additional infrastructure is going to be
taken enabling a repeatable evaluation repeatability i have already mentioned to you controlling
the environment pre detection of issues affecting performance pre detection means even before
actually the cloud is deployed you want to pre detect what are the issues that are going
to be there which would be affecting the performance of the cloud environment and that has to be
done even before the deployment is going to take place so cloud simulator will help you
do that and designing of the countermeasures in the case of you know the performance degradation
due to some issues or you know certain things going wrong so what are the counter measures
that have to be taken and you know designing that and evaluating those aspects different
cloud simulators that are available include cloudsim which is probably one of the most
popular you know simulation platforms that is available in the community cloudanalyst
is another one third is green cla ah sorry greencloud ah next is icancloud the next one
is groudsim and the last one is dcsim in this particular list so these are some of the some
of the cloud simulators that are available for use if you want to simulate cloud before
actually deploying cloud so cloud sim is a simulation platform as i
said before which has different modules which has different classes for different cloud
computing environments including modules for data center you know for modeling data centers
modules for data center virtual machines applications users network topology so there are different
types of modules and their corresponding models that are made available it is cloud sim is
based on a java based environment it is written on a java based environment that allows to
examine the performance of application services it is also possible to dynamically add and
remove resources in cloudsim and cloudsim was developed at the university of melbourne
by the team led by professor rajkumar buyya he is very much popular in cloud computing
ah and in his lab the clouds lab of the university of melbourne this cloud sim platform was developed
the advantages of cloud sim include taking care of time effectiveness so cloud based
applications implementation has to be done in minimum time with minimum effort so time
effectiveness is one second thing is dealing with fix flexibility and applicability supporting
you know the use of diverse cloud environments enabling the modeling of application surfaces
in any environment these are the some of the advantages of cloud sim features of cloud
sim include i will just read from this list and because these are quite self explanatory
i would simply lead them without actually needing to expend them in further detail so
features include various cloud computing data centers different data center network topologies
message passing applications virtualization of server hosts allocation of virtual machines
user defined policies for allocation of host resources to virtual machines energy aware
computational resources dynamic addition or removal of simulation components and stop
stop and resumption of simulation so these are some of the features that are
supported by cloud sim in the cloud sim architecture we have different layers the topmost layer
is the user code layer which basically presents the different machine and application specifications
the middle layer is actually the cloud sim layer which provides the actual cloud environment
and also enables modeling and simulation of cloud and the bottom of most layer is known
as the core engine core simulation engine layer and these basically takes care of event
scheduling event scheduling means because we are dealing with discrete event simulation
different events like the creation of the virtual machine the porting of the virtual
machine on the host and so on so these are like discrete different events so scheduling
of the events is taken care of at the bottommost layer the core engine layer creating these
different entities these different virtual machines the data centers etcetera entity
creation is also done at the core engine layer interaction between the different components
the clock management because you know there is actually a clock class from which you know
the the whatever you know entities you ah you borrow you inherit you know you have to
take help of the clock that means the time ah you will be required to be inherited so
that these different entities are time synchronized and overall we have a clock managed solution
the top layer basically has different entities such as the users the physical machine the
virtual machines the applications and surfaces and scheduling policies so basically i will
show you with the help of this particular figure so what will happens is you have this
user code and then you have the simulation specification and the scheduling policies
so simulation ah simulation specification and the schedu ah you know ah the scheduling
policy again breaks into two parts one is the application configuration which has ah
sorry as a application configuration the cloud scenario and the user requirement so this
is basically that simulation specification so this is taking care of by these three different
components the cloud scenario user requirement and application configuration the scheduling
policy is taken care of by this subcomponents which is basically the user broker and the
data center broker the middle layer is basically ah the cloud sim layer which takes care of
the creation and simulation of dedicated management interfaces issues such as memory management
storage bandwidth and virtual machine creation and simulation are taken care of at this layer
this particular layer helps in solving issues like host provisioning to virtual machines
application execution management and dynamic system state monitoring it allows a cloud
service provider to implement customized strategies evaluating the efficiency of different policies
in virtual machine provisioning here is the overall cloud sim architecture so we have
different layers in cloud sim we have the user interface structure the virtual machine
services cloud services cloud resources and network so these are the different components
of the cloud same so user interface structure has two different
sub components the cloudlet and the virtual machine so cloudlet is sort of like a small
physical server which which will put the you know which will create these virtual machines
then we have what the virtual machine services constituting the cloudlet execution execution
of the cloudlet that is created and the virtual machine management then we have the cloud
services comprising of virtual machine provisioning bandwidth allocation cpu allocation memory
allocation and storage allocation then we have the fourth one which is the cloud resources
ah component of the layer which includes event handling sensor management data center and
the cloud coordinator and then we have the network layer which takes care of the network
topology and message delay calculation so the that was the cloud sim another production
another system which works on top of clouds sim is the cloud analyst which is a product
that was again developed by the group led by professor rajkumar buyya
so the crowd analyst is a simulation tool designed on top of cloud seemed to provide
a graphical user interface that will support geographically distributed large scale cloud
applications the overall purpose of cloud analyst is to study the behavior of such applications
under different deployment configurations so basically this cloud analyst will have
different metrics to trick take care of the performance of this cloud the different parts
of the cloud and so on the features of cloud analyst include it is
easy to use due to the availability of gui it has high level of configurability it has
a feature of flexibility of adding different components repeatability of experiments graphical
output provisioning with the help of charts and tables and easy extend extensibility with
the help of java swing and different other technologies the cloud analyst architecture
is given in front of you in the figure so we have two main components one is the the
cloud sim extensions by offering different metrics different you know different extensions
that will help you to analyze what is going on how the cloud is behaving and so on and
the graphical user interface so this interaction with the core cloud sim and the different
components the two components of cloud analyst are shown in this particular figure
so the cloud analyst list basically comes with different components such as the gui
package for front end development the simulation ah ah simulation component which basically
creates executes and holds virtual machines then we have the user base for user traffic
generation data center controller internet for internet working routing networking network
provisioning and so on internet characteristics component which basically takes care of properties
of the internet with respect to delay bandwidth throughput etcetera the vm load balancer which
takes care of issues of policies for load balancing and the cloud app service broker
which takes care of entities for which has entities for routing between the user base
and the data center so why do we need green cloud so sorry oh
so that was the cloud analyst and now we have another cloud simulation platform which is
known as the green cloud so why do we need green cloud so in green cloud so by the way
this green cloud was developed by a team ah of us and europe ah pascal bovry group and
semi olokun you know from ah the university of north dakota and pascal bouvry from university
of luxembourg in europe so they together came up with discrete green cloud platform which
is again a packet level simulator which is energy aware and that basically helps in ah
reducing the overall you know energy expenses ah with the help of ah energy expenses in
the adoption of cloud so why do we need green cloud the computing
capacity has increased the cost and operational expenses of data centers so energy consumption
by data center is the major factor that drives the operational expenses so what is green
cloud it offers operational cost so operational cost is the energy utilized by computing and
combination communication units within a data center and how that is done green cloud basically
monitors the energy consumption of the servers the switches etcetera and it is developed
as an extension of ns two packet level network level network simulator
the features of green cloud are listed over here i would just read read it out for your
convenience so number one feature is that it is it offers a user friendly graphical
interface number two is it is open source next is the facility for monitoring energy
consumption of network and devices forth is that it supports simulation of cloud network
components fifth is it supports monitoring of energy consumption of individual components
next is that enables improved power management schemes which is very much important in the
context of this particular simulator energy consumption monitoring and reduction is an
important feature that is offered by green cloud and the last one last feature is dynamic
management and configuration of devices so those were the simulators now let us talk
about some real commercial and open source cloud platforms that are available for use
so some examples of opens open source cloud include openstack cloudstack eucalyptus and
so on commercial cloud platforms include amazon web services microsoft azure google app engine
and so on so now let us look at the open source cloud platforms these open source cloud platforms
mostly offer infrastructure as a service whereas the commercial ones offer in addition to infrastructure
as a service platform is a service software is a service and so on on a subscription basis
or that means on a payment basis in terms of security these security issues are implemented
by the user that means by the customer at the user end and commercial clouds these are
implemented the security aspects are implemented the security components are implemented by
the service provider so here the type of the cloud is private on
premise that means in the facility of the user the cloud will be installed and so on
on premise and the commercial clouds are basically public of premised so these are not available
on the campus on the in the institutional premises and so on so these have to be available
made a these are made available to the public and these have to be subscribed to over the
internet so openstack is one very popular form of a cloud platform ah that can be used
so openstack is a collection of open source technologies that is managed by the openstack
foundation it supports vastly scalable cloud system it has preconfigured software units
different services are available for the users it considers infrastructure as a service so
openstack basically supports ias and not a saas or paas it is easy to use because you
know one can easily add new instances and can ah quickly run different cloud components
it provides a platform to create software applications and has been developed it has
developed software applications which can be used by the end users
so this is basically the schematic of openstack so we have the common network component which
includes a container a storage and virtual machine so this is basically the common network
component then you have these different apps at the users apps layer and then it also connects
with the dashboard geographical interface and the different monitoring tools are also
available through the openstack platform openstack has different components and features i am
not going to go through but as you can see over here they have different components for
computation networking storage you know and so on and so for database etcetera etcetera
and they have different names for each of them nova neutron cinder glands keystone swift
horizon trove etcetera etcetera these different components have their own different names
the features include allowing users to create and deploy virtual machines allowing the setting
up of cloud management environment supporting easy horizontal scaling that means dynamic
addition and removal of instances to support more users in real time and open source software
this is open source software which is freely accessible to anyone we along with the source
code which can be shared to ah shared in order to ah share with the community for the deployment
and use of this platform microsoft azure is available on a payment
basis it is not free it ah you know earlier it actually it used to be known as the windows
azure it supports infrastructure as a service and as and also the platform as a service
so infrastructure as a service was also offered by openstack but not platform as a service
so this paid software microsoft azure basically comes with the platform as a service in addition
to the infrastructure as a service it supports extensive set of surfaces to quickly create
deploy and manage applications there are many many programming language support and frameworks
that are available in this particular ah platform the azure platform and it is available across
a worldwide microsoft managed data centers so these are the different advantages of microsoft
azure and here the list of different services that are supported by as your are given in
front of you we have support for computing support for mobile services storage services
data management messaging media services content delivery content delivery means that you know
something like offering different types of media ah you know etcetera through platforms
like youtube and so on so content delivery network ah del developer
ah ah you know developer services management services and machine learning support as your
as a platform as a surface paas platform is provided to clients to develop and deploy
software so this is very important so when we talk about platform as a service we are
talking about some platform that is offered for the development of the software to the
clients you know clients get the development platform for use on a ah after makings the
payments so clients basically focus on application
development rather than worrying about the hardware and the infrastructure as your is
low cost is less vulnerable to security attacks as claimed by them then it is easy to move
on to new tools with the help of azure it also solves the issues related to most of
the operating system servers and networks azure as infrastructure as a service the previous
one was platform as a service and the next one is infrastructure as a service ah in this
particular module it offers total control of the operating system and the application
stack it has features to access manage and monitor the data centers and it is ideal for
the application where complete control is required
the next one which is also quite popular is the ec two amazon platform ec two elastic
compute cloud the name says it actually the advantages of amazon ec two is evident from
the name so a web service for users to launch ec two is a web service for users to launch
and manage the server instances in amazons datacenter it provides various apis tools
and utilities it has the facility for dynamic computation scaling in the aws cloud and it
supports paper use billing rather than making large and expensive hardware purchases so
amazon has different ah is ec two amazon ec two has different instances so these instances
are of different types for different serving different purposes and their specific names
that are given by them in ec two are also listed over here
so in terms of the operating system ec two basically supports all operating systems ah
in terms of storage it has temporary storage for local instance ah ah store which is a
local instance store and also the amazon elastic block store the ebs and the third is the simple
storage service s three so these are the persistent storage mechanisms that are available to a
in ec two ah then the automated scaling so which is basically for horizontal scaling
where there are rules and schedules that are given and the scaling is going to be on the
basis of that there are different available zones in the data centers that basically increases
ah ah fault tolerance so ec two basically comes with the concept
of available zones availability zones so in the availability zone basically it is designed
the zones are designed in such a way that if there is some component that goes down
there will be some other component that is going to automatically take over so the zones
basically at the time of creation basically ensures that such availability is there so
that basically improves the overall fault tolerance in data centers
features of ah ah amazon ec two also include firewall support for firewall rules and securitys
overall ah there is predefined protocol ports etcetera ah which has source ip ranges ah
supporting different firewall rules and security mechanisms elastic ip address mapping which
basically maps between the ip and the vm of users so what you have essentially is one
pool which is basically the ip address pool the other one is the pool of virtual machines
that can be created for the user so basically this elastic ip addressing will map between
the ip addresses and the virtual machine ah ah and the corresponding virtual machine
amazon cloudwatch cpu disk network resource utilization monitoring these are some of these
functions of cloudwatch then you have the enhanced security mechanisms as features in
ec two and the last one that i would like to mention specifically is the formation you
know is the availability of feature for creation of virtual private clouds so which will basically
logically separate the private clouds not private cloud but it it is sort of like a
virtual private cloud so i can install in my lab a virtual private cloud which will
be logically separating ah from the rest of the amazon web services cloud and this can
be optionally connected to the users own network so with this we come to the conclusion of
this lecture we have gone through two different aspects of cloud one is the different simulation
platforms they are there clouds sim cloud analyst so on and so forth thereafter we spoke
about the actual ah you know cloud platforms that are available for real deployment so
in which again there are two classes of such kind of tools one is the open source once
so ah openstack by open foundation is openstack foundation is a an example of a open source
ah cloud platform the paid ones include amazon ec two amazon web services microsoft ah microsoft
azure and so on so with this we ah you know stopped over here with the cloud and ah also
in another lecture you will be given some hands on demos some demos with which you can
perform some hands on ah ah you know experimentation with cloud if you have the facilities with
you you can you know when we go through these ah different you know steps ah in the next
lecture you can also perform the experiments yourselves with the adequate facilities at
your end thank you

Leave a Reply

Your email address will not be published. Required fields are marked *