Don't let complexity affect cloud service promotion

Complexity is a potential killer on the road to cloud computing: the cloud-based system architecture becomes more complex in order to achieve the scalability and reliability promised by cloud services. The service itself must be designed with sufficient redundancy and self-monitoring capabilities, data must be automatically backed up to multiple different locations, and workloads must be balanced across multiple servers. As more and more critical applications migrate to the cloud, the risk of system errors, as well as cost losses associated with any errors, will increase accordingly.

In traditional IT systems, the way to minimize risk is to set up monitoring and alarm systems. Unlike these systems, as cloud computing systems become more complex, no one or team can respond adequately and in a timely manner to the failure of the system. As the interaction between cloud services becomes more and more strong, it is difficult to quickly lock down the starting point of System errors, and manual intervention investigation often leads to greater risk.

Hiring a large IT team to manage a cloud service system means that the cost savings of migrating enterprise applications to cloud services are completely lost. Reducing the complexity of cloud service systems does not seem like a viable route.

and the only viable solution is to rely on automated task management tools. With the evolution of cloud architectures, platforms, and software, a number of professional cloud services management solutions have emerged that will allow businesses to live more comfortably on the cloud.

Cloud based monitoring and alerting

The best products in such tools are cloud-based automated monitoring tools. Such tools enable anyone to monitor multiple resources on different servers and set up alarm triggers based on any parameters that are monitored. Such products are represented by Amazon's Cloudwatch and Cloudkick (which has been Rackspace acquired). These two solutions all support real-time monitoring of resources, with a variety of alarm settings. These two products also have a rich visual interface and scalability. Cloudkick supports so-called "Plug-ins", which are custom-defined monitoring scripts, while Cloudwatch communicates with other applications via API to obtain monitoring information.

The difference between these two solutions is the scope of their monitoring. Cloudwatch's monitoring focuses on Amazon's own servers and services, and supports other servers through APIs and the corresponding development of customers. While Cloudkick can support more cloud service providers, although this support is implemented through proxies, these agents can support multiple operating systems. So he can quickly and easily start monitoring any cloud server, regardless of which cloud service provider.

Another scenario is cloud server Management services provided by many large cloud service providers. Although this scenario is not a true automation tool, it makes it easier to manage cloud servers. Unfortunately, these services are generally very expensive. And if price differentials are taken into account, another option for high reliability is to select multiple cloud service providers, which means that there will be different people to manage different parts of the enterprise architecture (and basically different service levels), which is largely unattractive.

From monitoring to self healing to more advanced functionality

Once the above monitoring solution has been applied, the tasks that were originally handled manually by the administrator are automatically performed. Many cloud service providers have their own alarm systems that can be integrated into a monitoring solution that can send alerts to administrators via e-mail, telephone or SMS messages. But the alarm is only the first step.

Vendors such as Rackspace and Amazon can provide APIs that enable a series of automated tasks to be implemented with a reliable system. For example, once the monitoring system detects CPU or memory overload, the system will automatically increase CPU or memory processing capacity, while the alarm information passed to the technical support team. In extreme cases, the scheme can also automatically line up the problem server and dynamically assign the server's IP to the new server, which is imperceptible to the user.

Monitoring automation is a double-edged sword

Like many technology based solutions, monitoring automation solutions are a double-edged sword. In some failed cases, automated data recovery processes can lead to catastrophic consequences. In fact, Amazon has seen this in the last 4.21 days. A configuration error at that time resulted in a usability error, which in turn triggered the system's automated response mechanism, which eventually led to a wide range of systems being unavailable. At the same time, a series of elements that must be determined during the setup process make it very impractical for human to participate in management.

Of course, if you rely on the traditional monitoring system to track cloud application services, it is not not, but the traditional monitoring system is not designed for monitoring cloud applications. Most of these products are either not commercially developed for the cloud environment or are missing some of the functionality required to monitor cloud services, or the license fees are too high. Automating and rationalizing the use of APIs is important for monitoring systems, especially in the cloud environment. Therefore, enterprises should use professional cloud solutions as far as possible.

(Responsible editor: The good of the Legacy)