- Internal Code :
- Subject Code :
- University :
- Subject Name : IT Computer Science
Toward Cloud Computing Evolution: Efficiency vs Trendy vs Security
1. No doubt that Distributive system is a novel technology which opens a new dimension in the field of computation, but it creates many new issue and challenges. Computer researchers are working rigorously to addresses the existing issues of distributive computing. This section of the report provides emphasis on some of the important issues of distributive computing like security, configuration, performances, resiliency etc.
Security: It is the prime concern for any distributive system. Unlike traditional centralized systems, security threats are very high for the distributive computing platform. The main reason behind this is the use of network connection for sharing of the computing information among multiple nodes of the computer system (Lastovetsky, 2012).
Configuration: By definition, a distributive system supports a heterogeneous computing system (hardware) that runs some common program and follows common protocols. Due to the heterogeneous nature of the underlying hardware, it is difficult to configure a distributive system.
Performances: Distributive system uses multiple computing devices for performing any computation in a distributive manner. So its quite oblivious that the distributive system helps to improve the performance of the system. Now the question is that if a distributive system has N nodes that will it work N time fester that a centralized system that has one node which has the same capability as the nodes of the distributive system? The answer to this question will be no. If we consider that the target algorithm is highly distributive in nature, there will still be some overhead for the task decomposition, distribution and combination processes. Moreover, every algorithm is not fit well in the distributive system. For this reason, the improvement of the performance may not grows linearly for the distributive system.
Salability: Theoretically, we may increase the number of nodes (horizontal scaling) in a distributive system without any bound. But in the real-life scenario, if we incise the number of nodes after a certain threshold, then the system may suffer in performance issues. This is happening due to the coordination overhead. Apart from this, adding resources to the distributive system without taking any downtime is also very challenging.
Resiliency factor: This is one important factor of the distributive system. This parameter defines fault tolerance and capability to recover from the fault in quick succession. This parameter does not want to make a system fault free rather tries to minimize the impact of the fault and graceful restoration. As compared to the standalone system, achieving Resiliency is quite difficult in the Distributive system as every node has a partial view and partial data of the system. Moreover, identification of the fault is also very challenging, though the distributive system at least solves the single point failure problem of the centralized system.
2. In a single word, it can be said that Every cloud system is a distributive system but every distributive system is not a cloud system. So in simple word, cloud systems are a proper subset of the distributive system. Theoretically, distributive systems are a large class of computing platforms where multiple computing devices (nodes) work together to achieve some collaborative goals. This computing platform, basically achieved the property of the pipelining in multiple devices connected to the network. As multiple nodes work together parallelly, so we can achieve a higher degree of computation. In a traditional centralized system, it is impossible to perform vertical scaleup limitlessly, so the distributive system was proposed. This also helps to eradicate the single point failure problem of the centralized system.
A distributive system may have several issues like resource scaleup, transparency, configuration factor etc. To address most of these issues, the researcher proposed a top-level abstraction over traditional computing. This approach just creates a common software layer on the top of a distributive system. This software layer allows the administrator to run multiple hypervisors parallelly but in an isolated manner. Each hypervisor can run client OS/ Software. Moreover, computing resources are also virtualized from the perspective of the hypervisor. So, alteration of the resource can be done on the fly by making simple provisioning (Susanto et al., 2012). Due to this reason Cloud platform able to support several clients together on the same platform. As every client-side application runs in a separate virtual machine, so the client got total isolation. Moreover, scale-up processes are very easy as every resource is virtualized is nature.
So from this discussion, it is very clear that cloud systems are an extension of the distributive system. The low-level architecture of the cloud is exactly analogous to the distributive system but the top-level abstraction of the cloud system helps to achieve vitalization in computation.
3. The basic concept of cloud computing is more or less the same in every existing cloud system, but based on the service provided by the cloud provides, there are mainly three categories of the cloud system. These three classes of the public proud infrastructure are IaaS (Infrastructure as a Service), PaaS (Platform as a Service) and SaaS (Software as a Service) (Sala-Zárate & Colombo-Mendoza, 2012). Basic details of these three classes and their comparative analysis is shown in the following table.
|
Type |
IaaS |
PaaS |
SaaS |
|
Full form |
Infrastructure as a Service |
Platform as a Service |
Software as a Service |
|
Definition |
Cloud service provides only provides virtualized computing resources to subscribers. Subscribers can install any OS or software to this platform. |
In PaaS based service, the owner of the platform provides a platform (it could be an OS or development environment) to the subscriber. Subscribers can run suitable applications on that platform but can not change the platform. |
In SaaS-based services, the owner of the cloud provides an instance of a particular piece of software for the subscriber. End-user only allowed to use that particular software as a part of cloud service. |
|
Example |
Amazon Web Services (AWS), DigitalOcean, Google Compute Engine (GCE) etc. |
Google App Engine, OpenShift |
Dropbox, Cisco WebEx |
|
Level of user’s control |
The level of user’s control is over the platform is very high as the user can select the OS, middleware, runtime and application. |
The level of user’s control is moderate. User can control the data and the application |
The level of user’s control is very low. Users are allowed to control the data only. |
|
Data privacy |
Data privacy is high in IssS as users have the most of the control over the system. |
Data privacy is moderate as user’s control over the OS is limited. |
Data privacy entirely depends on the service provider. Users have very limited control over the data privacy parameter. |
|
Advantage |
The flexibility level is highest in the IaaS. The level of automation is high for the server deployment, processing power scaleup. Resource allocation can be done on the fly. Resource cab be purchased Just in Time manner. Level of scalability is highest |
It is similar to the IaaS but here the cloud provides directly provide the operational platform. It takes less time in the configuration. Less technical expertise required to manage the platform by the client. Level of complexity is less in PaaS |
Help to reduce cost of the software purchasing. Save time for software configuration, management, upgradation. No need to hire technical staff to manage the system. Save memory requirements in the client-side. |
|
Disadvantage |
The configuration technique is difficult and time-consuming. A technical expert is required for the clients of the platform. |
It is not possible to run every application on a PaaS platform. We can run only those software / application which is compatible with the underlying platform. |
Suffer from compatibility issues with other systems. Data privacy is very low. Control over the platform is less for the end-users. |
References for Cloud Computing on Distributive System
Lastovetsky, A. (2012). Special issue of journal of parallel and distributed computing: heterogeneity in parallel and distributed computing. Journal Of Parallel And Distributed Computing, 72(10), 1397. https://doi.org/10.1016/j.jpdc.2012.06.002
Sala-Zárate, M., & Colombo-Mendoza, L. (2012). Cloud computing: a review of paas, iaas, saas services and providers. Lámpsakos, (7), 47. https://doi.org/10.21501/21454086.844
Susanto, H., Almunawar, M., & Kang, C. (2012). Toward cloud computing evolution: efficiency vs trendy vs security. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.2039739
Remember, at the center of any academic work, lies clarity and evidence. Should you need further assistance, do look up to our Computer Science Assignment Help