Lade Inhalt...

Cloud Architecture. What are the possibilities of cloud computing?

Hausarbeit 2019 27 Seiten

Informatik - Sonstiges


Table of Contents

1 Introduction

2 Basics of Cloud Computing
2.1 Service Models of Cloud Computing
2.2 Public, Private and Hybrid Clouds
2.2.1 Public Cloud
2.2.2 Private Cloud
2.2.3 Hybrid Cloud

3 Cloud Technologies
3.1 Virtualization
3.2 VMware ESXi
3.3 Containerization

4 Scenarios of Cloud Computing
4.1 Netflix
4.2 Salesforce

5 Implementation of a Cloud Architecture
5.1 OpenStack
5.2 OpenNebula
5.3 Comparison
5.3.1 Basics
5.3.2 Administration
5.3.3 Popularity and Distribution
5.3.4 Architecture
5.3.5 Hypervisors
5.3.6 Integration of ESXi
5.4 Integration of Docker

6 Conclusion

7 List of figures

8 Bibliography

1 Introduction

The cloud and cloud computing are topics everyone is currently talking about. It is often used as a buzzword in marketing to improve the sales just because it sounds modern. Nevertheless, the cloud is a very important topic in the IT environment nowadays. There are probably still a lot of companies that do not use cloud tools at all or just a small portion of the benefits, simply because they do not know what is possible and what is not possible.

This paper is going to explore the possibilities of cloud computing, which effects different cloud architectures have and in which scenarios they are relevant. First, the basics of cloud computing are explained, how it basically works, and what the standard cloud models are. Afterwards different scenarios for cloud computing and their technological architectures are shown. To display how different cloud architectures can be build, the solutions of OpenStack and OpenNebula for creating a cloud infrastructure are introduced and compared to each other. To frame the paper, the topic will be shortly summarized in the end.

2 Basics of Cloud Computing

Cloud computing is basically serving computing resources of different levels of abstraction, most of the times over the internet using a pay-per-use model. The resources can be concrete resources like storage or networks, but also for example web applications. The target of cloud computing is to reduce the configuration and management expenditure for end users and organizations. Customers of a cloud system have none or only very little setup requirements like installing software. Computational resources should be rapidly scalable, so that usage spikes can be handled with ease and often the resources of a cloud system seem unlimited. Furthermore, concrete resources like storage or processing power should be dynamically assignable, independent of the customer’s location (see Brockhaus 2018 and Mell/Grance 2011: 2).

2.1 Service Models of Cloud Computing

There are three common service models of cloud computing.

First, there is Software as a Service (SaaS). When some piece of software is available over the internet, directly targeting the end user, without the need to install anything, it is called Software as a Service. It is usually accessible through various devices e.g. via a web browser. A Software as a Service can consist of one or more small components that can be integrated and used by other applications, e.g. OpenID, or it can be a complete application, like Google Maps or Google Docs (see Baun et al. 2011: 37).

Second, there is Platform as a Service (PaaS). A Platform as a Service is usually not for end users but for software developers. The platform can offer an environment for the developers to program their applications, or it can offer a runtime environment on which the developers can deploy their own applications. A Platform as a Service aims for easy and fast setup, installation and deployment. Customers do not have access to the underlying infrastructure like servers or networks, but they have the complete control over the deployed applications. Some popular examples for Platform as a Service solutions are Google App Engine and Microsoft Azure (see Baun et al. 2011: 35 and Mell/Grance 2011: 2 f.).

Last, there is Infrastructure as a Service (IaaS). An Infrastructure as a Service solution offers an abstract view on hardware, e.g. computing power, storage or networks. The cloud provider offers a user interface to manage all the needed resources. Creation or deletion of operation system images, capacity scaling for computation power or storage, or the definition of network topologies are common use cases. Moreover, the interfaces offer operative functionality for starting and stopping concrete server instances. Depending on the provider, the actual infrastructure can consist of concrete physical hardware, or it can be virtual resources, based on virtualization or containerization. Even though some providers offer a whole range of services, they usually have one main service or technology, e.g. Hadoop MapReduce is used for calculations, Dropbox offers storage, or OpenFlow is used for networking (see Baun et al. 2011: 32 ff.).

An overview of the cloud service models can be seen in Figure 1.

Abbildung in dieser Leseprobe nicht enthalten

Figure 1 : Cloud computing service models (adapted from Schouten 2013)

2.2 Public, Private and Hybrid Clouds

Besides the comparison of the cloud service models, cloud computing systems should be distinguished by their availability to the general public and their distribution.

2.2.1 Public Cloud

If a cloud is available to the public and everyone has the possibility to access it, it is called a public cloud. Public clouds are operated by third party organizations which run and maintain the whole infrastructure. The cloud resources are usually available for any user over the internet. The pricing model is usually pay-per-use for very short time frames, so the users can allocate and free the resources on the fly. This way, they can easily scale their environments and make sure that their services are available all the time. Additionally, companies can reduce their costs for server, as there is no need for additional backup server that are only used at peak times. As the third party vendor is responsible for maintaining the servers, the availability and uptime is usually very high and stable (see Goyal 2014: 23). Examples for public clouds providers are Amazon Web Services (AWS), Microsoft Azure or Google Cloud Platform.

The main drawbacks of public clouds are data privacy and security. When data is stored in a public cloud, an organization may not know where the data is stored exactly, how it is backed up, and how secure the data is in terms of accessibility by other people (see Goyal 2014: 23). Especially today, with the General Data Protection Regulation (GDPR) in force, one has to be very careful to make sure that the public cloud is GDPR-compliant when storing sensitive data in the cloud (see Roßnagel et al. 2018: 294). As public clouds are open to everyone, they usually do not reflect the security needs of larger companies, even though they are quite secure in general. Therefore, the usage of a public cloud should be planned very carefully to make sure that the cloud meets all the requirements and does not lead to a new threat for the company. Nevertheless, in smaller companies the security issues can often be considered a benefit instead of a drawback. As they often do not have enough manpower to handle every aspect of security, they could use the public cloud, so they do not have to worry about security for e.g. the data storage (see Goyal 2014: 23).

2.2.2 Private Cloud

A private cloud is, opposed to the public cloud, a cloud that is only available for a single organization or a very small group of connected companies. The functionalities and the data that are made available through the cloud are only used internally, for example to share logistics data between company locations. The private cloud can be operated by the company itself or by a third party organization. Geographically, it is possible to either host it directly at the company, e.g. in their data-center, or at another place, which is less common than the former (see Goyal 2014: 23).

Generally, the usage of a private cloud is not different to the usage of a public cloud. It is also self-service-based, so you can get resources on-demand and the infrastructure is automatically managed by the cloud system. A private cloud is especially interesting if a company has otherwise unused capacities. There might be some servers with very high load and others that idle most of the time. In this scenario, the cloud management software could redistribute the capacities to balance the workload. Another very important advantage is the aspect of security and data privacy. Opposed to the public cloud, a company has all the data as well as the security configurations in their own hands, so they can exactly tell, how secure the data is and where it should be stored (see Goyal 2014: 23).

Even though the private cloud offers a lot of advantages and reduced costs compared to the regular data-centers, it still has disadvantages. Mainly, it is in general more costly than the public cloud as there are the costs for hardware, software, and staff. Furthermore, it is not as flexible as a public cloud. New hardware is required whenever the required calculation power exceeds the available, and there are additional servers to handle traffic spikes (see Goyal 2014: 23).

2.2.3 Hybrid Cloud

Hybrid Clouds combine the advantages of both private and public clouds. One or more private clouds are combined with one or more public clouds to a single, unified system. The clouds are connected via a software system that offers portability of data and functionalities, so the systems can be interconnected. The hybrid cloud is thus usually much more complex than a single private or public cloud (see Goyal 2014: 24).

In a hybrid cloud, some data can be stored inhouse in the private cloud. Usually that would be data which needs high security standards as private data of end users. On the other side, data which does not need specific security treatment can be stored in the public cloud, e.g. data for general statistics. Meanwhile, it is possible to maintain the scalability and cost-effectiveness of a public cloud. When the hybrid cloud system experiences usages spikes, there is no need for additional servers as a buffer, as the overhead can simply be overcome by outsourcing some of the traffic to the public cloud. After the spike, the system just removes the public cloud resources, so the additional costs are minimized (see Goyal 2014: 24).

The drawbacks of hybrid clouds are the complexity, and therefore also security issues. A hybrid cloud is not as easy to set up and maintain as a single public or private cloud. It has to be configured and managed carefully to prevent security risks. Additionally, the general security issues of the public cloud are also present. The communication channel between the public and private cloud offers an extended surface for attackers. Hybrid clouds lead to a larger infrastructure and thus more tools in use. These are interesting targets for attackers. For example the hybrid cloud management tool which communicates with all the systems, as it has many external interfaces. Nevertheless, when being implemented thoughtfully, hybrid clouds offer great opportunities and flexibility for different purposes, as there is no standard solution for everyone (see Goyal 2014: 25).

3 Cloud Technologies

Users of cloud services want to make use of computation power, storage, and networking resources. The resources have to be split up dynamically between the participants without disturbing each other. The users want to obtain resources on the fly and at any scale. To cope with this problem, there are different technologies which are going to be discussed in this chapter.

3.1 Virtualization

The major technology used today to handle cloud infrastructures is virtualization. Virtualized environments hide how resources are divided between the servers by offering an abstract view on the overall resources. For consumers, the resources seem to be infinite (see Fox et al. 2009: 2).

Virtualization basically means splitting up one computer in multiple parts, so several operations systems or applications can be executed at the same time. To achieve this, one or more physical hardware systems can be emulated software-sided by one physical computer. It appears like there is more than one physical computer, each with potentially different operation systems and applications running (see Dziak 2018).



ISBN (eBook)
ISBN (Buch)
Institution / Hochschule
Fachhochschule Dortmund
cloud cloud architecture openstack opennebula cloud computing docker cloud integration cloud infrastructure



Titel: Cloud Architecture. What are the possibilities of cloud computing?