How to better manage the efficient cloud application

APM is a monitoring process that adjusts application resources to meet specific experience quality standard sets by applying business use cases. Technically, QoE is the combination of application execution time and network delivery time, and these can be diversified in the cloud. Successful Application Management (APM) in the cloud involves several key steps: setting resource boundaries temporarily, restricting performance variables, and applying monitoring practices to cloud situations, and the last one is deploying compensation measures when direct resources do not solve experience quality (QoE) problems.

In fact, one of the most important features of cloud computing is that when the pool of potential cloud resources is large and geographically different, the network response time will inevitably be different between resource pools in various locations. A far-away managed point usually arrives with more route hops, resulting in more latency, but the number of accurate hops between your users and cloud hosting points can be significantly different between potential cloud network providers.

Simple tests, for example, use the Routing Tracking Diagnostics tool to establish connectivity performance from each major employee location to different points in the cloud, helping to identify network providers with the best performance.

Monitor performance measurement response time

Once you have done everything you can to control the application performance variables related to the application of the managed point cloud distribution, the next step is to refactor the application supervisor

Control practice and tools.

Typically, APM begins to measure response time from the user level, and then "fall back to Application" through the connection and functional continuum layers. The APM tool can be applied to the user Service point or to the internal application/component itself, making it possible to deploy the same tools and practices for the cloud applications used by the data center.

The only requirement for cloud APM is that the expected tools coexist with applications/components that must be part of the deployment software mirror, which means that the cloud services hardware and software platforms must be compatible.

Some APM users deploy network probes or other network management tools, detect application packages at key points, isolate latency resources, and identify problems that are clearly not available in the public cloud. The only realistic monitoring strategy is that the detection packet can only be at the network boundary point, which means that the connection point connects to the user and the application component. It is likely that the APM tool has monitored user boundaries, so it may be necessary to consolidate network monitoring and application mirroring so that tools and applications can be deployed to the cloud and accessible.

Cloud services involve a number of operator-supplied connection points, which are difficult to implement, unless one or two operators provide a monitoring probe when connected. You may also find problems by tracking routes, but only when operators expose their infrastructure to control the protocols they use. If not, subsequent fault isolation and specific network remediation (through service level agreements) can be difficult.

The goal of isolating the source of the problem is to fix the performance problems caused by the specific problem-gravity-separation routing connections, changing hosting locations, and so on. When due to lack of resource control, it is not possible to isolate the problem or make the required changes, a compensating performance enhancement is needed to improve cloud APM.

Accelerated Cloud Application Performance

Effective APM technology is divided into two major groups: Network acceleration and component replication for load sharing. One of the biggest mistakes that the IT staff can make is that the former can be used for network problems, and the latter for computing problems. Any performance improvement can be used to compensate for enhanced performance, whether or not based on performance issues, because the goal is to remedy the problem.

Network performance enhancements typically involve data compression, multipath transmission, and traffic prioritization. About half of the companies in their applications use some form of network performance, so they naturally expect to be able to migrate the same tools to cloud APM.

The problem is that technology requires a pair of devices, one at each end of each network path, but it is not possible to place applications on the other side of the cloud. You want the network performance tool to operate server-side software, not a device. But make sure that the software is compatible with the hardware and software of the cloud, because the machine mirroring must be consolidated for deployment.

Application component replication provides additional parallel processing capacity to improve performance under load, but this mechanism can only be applied in the cloud when applying load-induced performance problems. If you doubt this situation, the best option is a higher-performance server or dedicated server.

However, if server sex solves the problem and does load-related, consider replication. For replication to work, the application must be designed to be able to run a set of parallel instances, with load balancers allocating work. In order to apply in the cloud, the load balancer may have to be a cloud-hosted software component.

Most cloud performance issues can be resolved by tuning cloud and network connectivity, following the same generic program used for private datacenter application hosting. The risk is that the program needs to remain qoe within the boundaries, resulting in additional cloud hosting costs for specific services that could jeopardize operational cases.

"Edit Recommendation"

Private Cloud performance monitoring: Developing aggressive planning cloud performance management tools: see through Cloud's eye private cloud performance problems fix "responsible editor: Xiao Yun TEL: (010) 68476606"