Common problems in cloud computing application deployment

The enterprise cannot complete the necessary upfront planning to determine which application is more suitable for cloud computing deployments.

Applications that require mainframe support, applications running on UNIX cluster services, and numerous existing applications running on mainframes are destined not to migrate to cloud computing. These programs are complex, and deployments based on the X86 hardware and software environment may require a redesign of the schema or rewrite of the application code. If deployed in public cloud computing, applications with high security requirements may not be the ideal implementation object for cloud computing deployments.

Companies cannot choose the right cloud computing model (private cloud computing or public cloud computing) to implement application deployment.

Applications can be deployed in private cloud computing or public cloud computing. Private cloud computing is the internal cloud computing that is under the control of the IT enterprise that created them. Compared with public cloud computing, they have more similarities to traditional data centers (they are internal, under the control of IT companies, without security issues in public cloud computing, and so on).

Public cloud computing is external. The infrastructure of public cloud computing is limited to cloud computing providers, and there may be many differences in their appearance and application experience compared to traditional data centers and even private internal cloud computing. A common mistake for an enterprise is the inability to determine whether the application is more suitable for public cloud computing or for private cloud computing. Another common mistake is the inability to determine the cost of deploying an application in each of the cloud computing models, whether long-term or short-term.

Businesses tend to focus on migrating servers to cloud computing rather than deploying applications in cloud computing.

When companies decide to implement migration from traditional data centers to private cloud computing, the motivation is often server consolidation, which can help improve server utilization and reduce capital and operational costs. These should not be the focus of attention. The real concern should be to deploy applications in cloud computing. By focusing on application deployment, organizations will have an in-depth understanding of the composition of applications and the management tools needed for applications in a cloud computing environment. This error causes some other common errors.

Inability to plan for application performance changes in cloud computing

Deploying an application in cloud computing can cause the application's performance level to be lower than its performance level in traditional data centers, due to differences between the two environments. When business executives consider the performance of an application, they typically focus on factors such as CPU processing power, memory capacity, and disk storage space. In a traditional datacenter, an application might run only on a single server. You can use the Physical Server monitoring tool to debug your application on that server so that the application can run at an acceptable level of performance.

When deploying applications in cloud computing, in a virtual environment created by Hypervisor software (such as Vmwareesxi or Xen), it shares physical CPU resources, physical memory capacity, and so on on a single virtual managed server with other applications. These applications compete for the physical resources of the virtual managed server. The performance tuning of applications in cloud computing begins with this new ecosystem.

Before an application is deployed in the cloud, you should create a baseline for the applications that can meet your business needs satisfactorily. When deploying an application in cloud computing, you should review its performance and compare it to baseline performance and make adjustments until it reaches the level of performance acceptable in cloud computing. For this type of performance analysis, you need to have a performance monitoring tool that works in a virtual environment.

Inability to understand the need for new tools to monitor application performance, security, and network traffic

Some enterprises do not understand the tools used in traditional physical environments and do not apply to virtual environments such as cloud computing. Monitoring tools can help answer the question: what is the performance of an application? Can an application access computing resources and storage bandwidth when needed? What is the response time for an application to access a storage device? Is my application protected from intruders?

Virtualization adds an abstraction layer to traditional monitoring. You can no longer monitor performance by looking at physical devices. The network operations team has been working to find such an abstraction to determine what is happening at the virtual and physical levels.

Because there is a lot of traffic in the hypervisor and not appearing on the physical network, you need to have tools designed specifically to run in a virtual environment. Physical layer based monitoring tools do not have the means to detect traffic between virtual components such as virtual servers, virtual routers, performance switches, and so on. Monitoring application performance and resource performance interacting with applications in cloud computing environments require new tools designed for virtual environments. The same thing happens in the area of application security. Tools such as Catbird Network company Vsecurity are designed to address security issues by monitoring virtual network traffic.

(Responsible editor: The good of the Legacy)