UPYUN launches CDN real-time performance monitoring

In 2014, entrepreneurs became the protagonists of the era. A large number of entrepreneurs emerged in the mobile APP, mobile games, O2O, Internet finance, online education and other industries, which brought the entire Internet into an era of data explosion. The overall CDN download acceleration of these data has become a problem that companies have to face. It is predicted that the domestic CDN market will exceed 5 billion in 2015, and the development scale of the CDN industry will further expand.

Cloud acceleration service provider UPYUN recently launched a CDN real-time status and performance monitoring service. When users log in to the UPYUN backend, they can intuitively obtain the "real-time performance display" and "real-time data statistics" of more than 120 CDN nodes across the country on the UPYUN platform under the current status. The real-time status of the CDN service is intuitively displayed in the form of maps and charts.

The purpose of UPYUN opening up its CDN real-time performance monitoring service to the public is to promote greater transparency in the current domestic CDN market. In the future, it may even provide customers with more comprehensive big data analysis services based on real-time performance and data monitoring to create more value.

By analyzing the data of more than 120 CDN node logs across the country, UPYUN has extracted accurate data in multiple dimensions such as "speed", "time consumption", "ISP line", "region", and "request share", and displayed them in the form of a map.

1. ISP lines, regions, and request ratios

If your end user sources include "Telecom", "Unicom" and "Mobile", then by switching "ISP Line", you can clearly see the performance of different lines. Among them, the data under the "Average" Tab is the average of all ISP line data.

When you move the mouse to a province, you can view the detailed data information of the province. As shown in the figure below, the currently selected line is "average". When the mouse moves to Hunan Province, you can see that the average speed of the telecom line is 516.89KB/s, the average time is 3312.61ms, and the request volume of Hunan Province accounts for 2.601% of the national request volume.

2. Speed ​​& Time

The table data on the right side of the map can show the speed and time consumption of each province in the country. The "-" symbol only means that there is no data request for the province in the previous minute. The data refresh frequency is one minute, and each refresh will be updated in real time on the national map and table data.

On the national map, different colors will be displayed according to the different data values ​​of "speed" and "time consumption". For "speed", red means slower speed and green means faster speed. "Time consumption" is just the opposite, red means more time-consuming and green means faster. Of course, the fast and slow here are relative values, which are only used to evaluate the performance of each province in the country within the previous minute.

It is recommended that the specific data of "speed" and "time consumption" be combined with "ISP line" and "request ratio" for reference, so as to have a more comprehensive understanding of the service quality of the entire CDN network.

3. Real-time data statistics

"Real-time data statistics" is similar to "Real-time performance display". It also uses a refresh frequency of one minute to count and display the service quality of the entire network in real time. It mainly includes four dimensions, namely:

Health: Count the HTTP Status Codes of all requests and calculate the proportion of 2xx and 3xx requests

Cache hit rate: the ratio of all hit requests to the total number of requests

Average download speed: Average download speed of all files

Average time consumed: The average time consumed by all files

Among them, after clicking "Health Details", you can view the specific proportion of all HTTP Status Codes, as shown in the following figure:

History is often like this, a seemingly small improvement can often change the development process of the entire world. UPYUN's move is likely to bring about a major change in the CDN industry. CDN*** has changed from a network acceleration tool to a service, and the service is to continuously meet various needs that exceed user expectations. The increasingly progressive needs of customers will inevitably drive CDN service providers to improve their service capabilities in all aspects more quickly. From the perspective of the development of the entire industry, this will be a good thing.

We also communicated with UPYUN CTO Huang Huipan about the technical implementation principle of UPYUN CDN real-time performance detection. He said:

Currently, UPYUN has deployed its self-developed second-generation traffic statistics system on more than 120 nodes and more than 2,000 servers in the entire network. It can collect statistics on more than 10 performance indicators such as traffic, number of requests, access speed and access status per domain name every minute (these data are generated during the user access process and truly reflect the user's access situation). These records will be transmitted to our data center within 1 minute, and the data sent from more than 2,000 servers will be summarized and analyzed in real time in the back-end data processing cluster to generate a complete data report stored in our UPYUN distributed structured storage system for the front-end page to call and read at any time.

If you are careful, you will find that the distributed structured storage system mentioned above, we UPYUN officially started research and development in this area in 2014, and it is currently mainly used for UPYUN internal systems, such as our CDN real-time performance monitoring service. The amount of data generated by the system every day is very large. For example, a domain name generates 1KB of data records per minute, which means 1440 records a day, totaling 1.4MB. UPYUN serves more than 100,000 customers and more than 200,000 access domain names, which means that the entire network will generate more than 2 billion records every day. We need to store these structured data persistently, and general relational databases can no longer meet our needs, so we began to develop a distributed structured storage system.

After our distributed structured storage system passes UPYUN's own business verification, we will also provide it to our customers in the form of a service.