Читать книгу Cloud Computing Solutions - Группа авторов - Страница 78

Cloud computing presents itself as a good servant to the end user, but it has some challenges and issues.

In the real-time system, it is too difficult to isolate the performance of the virtual machines, especially if the system is highly loaded. If the performance of virtual machines cannot be determined, cloud designers won’t be able to distribute the workload among the several data centers.

Cloud computing involves shared networks, fluctuations in bandwidth and various topologies. Due to these constraints, cloud computing infrastructure reveals latency while transferring the data from one virtual machine to another. Therefore, it affects the performance of the deployed applications.

Security issues are also issues of cloud computing. It’s very risky to put all the data of an organization in the hands of a cloud provider and running an application in another place is also at risky. A common problem is the threat of data loss due to a phishing attack. Therefore, privacy and reliability should be maintained during transportation and storage of the data. Cloud is made for sharing the workload in a common infrastructure and users have to rely on the cloud provider for their identity information, operation histories and perceptive data. Cloud is not responsible for unauthorized usage and its retrieval, lack of user administration and third-party access. From a piracy aspect, there is a pool of resources and millions of software applications and services being run publicly. So it’s very easy to pirate floating data from all those things and use it in an unauthorized way without any identification and authentication. It irritates a customer when CSPs promise to deliver a service but due to over-utilization are not able to meet their needs. Internet latency is also a drawback of cloud computing, which hampers the CSPs ability to deliver services on time. In regard to auditing, in a particular service or application, the CSP and the end user have the authoritative control over the data. CSP has the authority to replicate, shift and alter the data. That’s why the clients need to keep watch over all those activities so that the CSP can’t act beyond its domain. But it is not practical to audit all the data and also complicated to decide which data needs to be audited [1]. Moreover, multi-tenancy is also a problem if the number of applications running on a particular node is going to increase, in which case the bandwidth allocated to each application decreases, which means that the number of applications and allocated bandwidth is inversely proportional. This reduces the performance of the system.