5 Things To Know About Cloud Computing Performance
MikeEbbers 270001G1G2 Visits (5008)
Cloud computing is a way to dynamically allocate and deliver IT services in response to rapidly changing business needs without having to understand the technology itself. As the pace of business and society in general continues to accelerate, the physical and digital foundations on which progress depends are straining to keep up. Focusing on performance is key.
Here are 5 things to know about performance when using cloud computing.
1. Cloud environments often need to grow or shrink quickly.
Because cloud systems are built to handle changing workloads in a flexible manner, it is important to state our requirements and service levels so that the workloads can grow without impacting responsiveness, throughput, or availability. We must have a maximum workload level in mind, and the relationship between processing volumes and system resource utilization must be as linear as possible. Our key performance indicators must have corresponding thresholds that can alert us to possible issues.
2. Some workloads are better suited than others to cloud computing.
Those that contain asynchronous processes, for example, that can be offloaded to various areas in the cloud environment. The results are then combined and the application can continue. A web-based or high performance computing (HPC) scientific workload can also fit quite well, especially if there are computations or algorithms that can be offloaded and executed concurrently with extreme parallel processing. Workloads that are parallel and atomic take advantage of this environment to the fullest.
3. Because of the complexity of a cloud, performance testing must be properly designed and organized.
Generally speaking, cloud environments consist of a cloud management platform along with a managed platform. It is advisable to consider specific performance testing scenarios for each:
4. The objective of monitoring is to identify and address issues before consumers of the cloud are affected.
Monitoring should have dashboards or views for the cloud service provider and its consumers. Private and hybrid cloud consumers might also be interested in resource metrics to gauge if they need to increase their investment in cloud capacity and the external cloud vendor’s service qualities. Tip: use historical data and avoid online monitoring for this analysis.
5. Capacity management for a project running on a cloud is similar to any IT environment.
In addition to the two standard activities of establishing the framework and evaluating the process, there are four tasks:
We must build intelligent clouds that can handle increasingly diverse, variable, and challenging workloads while fully exploiting the emerging capabilities of new technology. This will allow us to deploy and manage cloud solutions that perform well while being cost-effective. For more information, see these two IBM Redpaper publications:
Mike Ebbers is an IBM Redbooks Project Leader. He works with technical experts to create books, guides, blogs, and videos. Follow Mike on Twitter at @MikeEbbers.