Shulou(Shulou.com)06/03 Report--

With the rapid development of operation and maintenance technology in recent years, most of the operation and maintenance teams have built a variety of systems: virtualization, containerization, continuous integration and so on. But how to effectively use these systems to achieve high availability, high performance and high scalability of the site? With the development of intelligent technology, in order to solve the above problems in the field of operation and maintenance, the voice of intelligent operation and maintenance is getting louder and louder. According to Rao Chenlin, product director of * * log easy * *, the development of domestic intelligent operation and maintenance is still in an exploratory stage, and if we want to make a breakthrough in the field of intelligent operation and maintenance as soon as possible, first of all, we should focus on the monitoring system and alarm system, and use machine learning algorithms for rapid monitoring and troubleshooting. Br/ > in the view of Rao Chenlin, product director of * * Log easy * *, the development of domestic intelligent operation and maintenance is still in an exploratory stage. In order to make a breakthrough in the field of intelligent operation and maintenance as soon as possible, we must first focus on the monitoring system and alarm system, and use machine learning algorithms for rapid monitoring and troubleshooting.

1. Views on the current situation in the field of domestic operation and maintenance

To put it simply, at present, the domestic operation and maintenance industry has reached a certain level of automation, at this stage, there is still a certain distance between automation and monitoring. For example, if you get a monitoring alarm, you may not be able to immediately know which automated deployment to do. If the two parts of automation and monitoring can be organically combined, the work of operation and maintenance will be more convenient.

2. Challenges to operation and maintenance brought by mobile and micro-services

In the era of PC, a lot of work of operation and maintenance is limited by browsers. When the operation and maintenance staff can not get the real data of the client, people will generally purchase some third-party services and try to obtain terminal data. In the mobile period, we all have our own client, we can put some logic of the collection point in our own client, and then get more accurate and real customer data. Compared to the challenge, this is actually the benefit of the mobile side.

The emergence of micro-services brings some problems to the operation and maintenance work. Before the emergence of micro-services, operators can troubleshoot problems in one or two or three machines. In the era of micro-services, these problems may occur anywhere in dozens or even hundreds of nodes of distributed systems. The start, stop and migration of a single service in the system is very convenient and frequent, and the respective data output becomes very chaotic, which means that the operation and maintenance staff have a lot of trouble to find problems, and intelligent operation and maintenance is needed to solve these problems.

3. In the era of big data, the relationship between intelligent operation and maintenance, data and automation operation and maintenance.

The ideal state of intelligent operation and maintenance is to organically combine the three major parts of operation and maintenance work (monitoring, management and fault location) by using some machine learning methods.

In the era of big data, intelligent operation and maintenance was based on big data. At present, it seems that if the operation and maintenance want to combine monitoring, management and fault location organically, it is inevitable to use intelligent algorithms, and the value of intelligent algorithms has a premise: intelligent algorithms need a lot of data to support.

Automated operation and maintenance has been in a state of benign development in recent years, including the automation of configuration management like Puppet and the automation of deployment like Docker. Further development is the need to integrate these three parts. At present, the way to integrate these three parts is to use the means of artificial intelligence to achieve a state of intelligent operation and maintenance.

4. the current situation of intelligent operation and maintenance and the prediction of the development of intelligent operation and maintenance.

At present, intelligent operation and maintenance is still a stage of preliminary exploration. I can give you a few time figures. One of the open source projects I saw related to intelligent operation and maintenance was in 2013. The first domestic enterprise to take the initiative to promote intelligent operation and maintenance was Baidu, which was in 2015. A large number of intelligent operation and maintenance appeared in the second half of 2016. These sermons are still in the experimental stage, and the effects of these attempts need more thought collisions to find ways to achieve better results, because we are now using ordinary machine learning algorithms, not algorithms like AlphaGo's deep neural network. The realization of intelligent operation and maintenance is a process that requires a lot of investment and learning.

If you want to make a breakthrough in the field of intelligent operation and maintenance as soon as possible, a more practical way is to focus on monitoring system and alarm system. Traditional IT operation and maintenance needs to manage a large number of alarms, which greatly distracts the attention of enterprises and consumes a lot of time and innovation of operation and maintenance personnel. To find a way to efficiently solve the problem of receiving thousands of alarm emails a day, and to extricate the operation and maintenance personnel from the numerous and complicated alarms and noise is a train of thought that can generate value quickly.

Now it is relatively clear that everyone will develop in the direction of intelligent operation and maintenance, and the development of intelligent operation and maintenance must be a long-term evolution process.

My simple views on the development forecast of intelligent operation and maintenance are as follows:

Intelligent operation and maintenance will first reflect its value in the alarm system.

In the first stage, we should judge the alarm intelligently, instead of relying on human experience to set a threshold. Setting an alarm threshold is a time-consuming and labor-consuming task, which can only be carried out by operation and maintenance personnel on the premise of fully understanding the business, and whether the business is developing smoothly. Otherwise, the operation and maintenance engineer will be absolutely exhausted if he changes it once a week or two.

In the second stage, the fault can be located intelligently. The current fault location completely depends on human experience and the completeness of CMDB, but maintaining a complex CMDB itself is a big problem. Intelligent operation and maintenance personnel should be able to assist operation and maintenance personnel to locate faults quickly from the aspects of correlation analysis.

In the third stage, some NLP (Natural language processing) technologies are used to deal with the fault reports written in natural language and automatically feed them back to the intelligent operation and maintenance system. This may be a more distant idea, but at present it will be a path for future development.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.