With the rapid growth of large-traffic services such as live broadcasting, video, games, and e-commerce in recent years, traditional CDNs are also facing increasing pressure to evolve. In addition to the initial basic functions such as “guidance” and “nearest neighbor”, “high availability” And “intelligence” has also become a problem that CDN acceleration needs to be optimized urgently.
Today Huosanyun Fusion CDN will compare two CDN scheduling methods for you: traditional CDN scheduling and HTTP-based CDN scheduling.
1. CDN Scheduling
①Scheduling based on DNS resolution
The traditional CDN solution uses DNS resolution for scheduling, uses the LocalDNS egress address as the query IP, and performs regional scheduling for users based on the query IP, but the obtained address may not be the real address of the user. CDN scheduling according to the wrong address obviously violates the original intention of CDN service “guidance” and “proximity”.
②Scheduling based on failover
A complete CDN system will have a failover mechanism for edge service nodes, and can meet the requirements of manual switching and automatic switching at the same time. But no matter what kind of failover, it is always a remedial measure after a failure occurs, which will cause fluctuations in the service quality of CDN acceleration and greatly affect the “available” service focus of CDN.
③Scheduling based on edge service node polling
By setting up an authoritative DNS server, the CDN service provider provides the corresponding edge cache service node A record set of a certain domain name in each region to LocalDNS for resolution query and cache.
There can be a maximum of 13 A record sets, and LocalDNS selects from the 13 A records and provides them to URL access users through polling, or provides all of them to Internet user clients, which are selected by the clients themselves.
The problem brought about by such a polling mechanism is: if the same Internet resource is stored in each service node, it may be necessary to go to the source site for 13 cache acquisitions.
The resulting access pressure, bandwidth cost, and subsequent poor user experience on the source site of Internet companies are also considerations for CDN service providers.
2. HTTP-based CDN scheduling
The HTTP scheduling system of the CDN includes a decision center server, one or more scheduling center servers, and one or more edge scheduling Cache servers.
The decision-making center obtains the global server cluster bandwidth and load conditions, and makes unified decisions to generate decision-making strategies globally.
The dispatching center implements the central decision-making strategy and establishes a direct connection with the user client. When receiving the client content request, it accurately matches the preferred server cluster according to the actual geographical location of the client and the content of the request, allowing the client to access through 302 jump , to obtain cached content and implement CDN acceleration services.
The HTTP scheduling system has the following characteristics compared with the CDN acceleration service:
① Scheduling based on HTTP protocol
The client and the server are directly connected through the HTTP protocol, and all client requests are directed to the dispatch center server.
The dispatching center server obtains the public network IP of the Internet user from the client’s request, and uses it as a benchmark for CDN scheduling to select the edge cache service node, which can accurately “guide” the request to the real “near” optimal edge service node.
②Scheduling based on bandwidth and load
The decision center of the HTTP scheduling system relies on the node collection system of the CDN network to obtain the real-time bandwidth information of the service nodes of the entire network of the CDN system and the real-time health coefficient of the cluster as load data.
The decision-making center coordinates the load data of the entire network, quickly calculates the decision-making strategy through an excellent decision-making algorithm, and provides the decision-making strategy to the dispatching center for execution at the first time.
The decision-making center can timely pull traffic and implement alternative service clusters for edge cache service nodes with high loads in the entire network, so as to prevent service nodes with high loads from resisting traffic and cause failures and downtime in advance, ensuring high availability of CDN services.
③Precise scheduling based on edge service nodes
Relying on the decision-making strategy of the decision-making center, the scheduling center can perform unique calculations on each Internet resource through precise algorithms, and determine the only edge cache service node to provide CDN acceleration services. In this way, multiple caches of an Internet resource at the edge cache service node are avoided, and the pressure and cost overhead brought by “caching from the source server” to the source site of the Internet enterprise are reduced.
3. Traditional CDN scheduling vs HTTP-based scheduling
From the above analysis, it can be seen that compared with traditional DNS resolution scheduling, HTTP protocol scheduling can guide Internet users’ requests to adjacent edge cache service nodes according to their real IP addresses, achieving absolutely effective CDN acceleration.
At the same time, since the HTTP server, that is, the scheduling center server, directly establishes a connection with the client, it can achieve precise scheduling at the stand-alone level, and make decision-making scheduling adjustments in real time based on current bandwidth information and cluster load conditions, greatly improving the quality of CDN acceleration services and customer satisfaction.
Furthermore, the flexibility of the HTTP server enables it to easily implement personalized acceleration strategies for different acceleration resources for enterprises that require Internet acceleration, which is also unmatched by traditional DNS scheduling.
Although the current HTTP-based CDN scheduling is still not perfect, it has broader usage scenarios, and it is very likely to replace the traditional DNS scheduling within the enterprise, realize a better CDN acceleration solution, and allow Internet users to get smoother online experience.
4. About Fire Umbrella Cloud Fusion CDN
Shenzhen Portal Technology Co., Ltd. was established in August 2014. It has obtained the domestic value-added telecommunications business license. The company adheres to the development path of technological innovation, and all core technologies are independently developed. It is a professional edge computing and integrated CDN service provider in China. The company’s main business includes integrated CDN services, game shield, server rental, etc.
Fire Umbrella Cloud Fusion CDN better solves the following problems:
Wide coverage of global intelligent integration
Integrating more than 30 mainstream CDN manufacturers in the world, users can freely choose up to 20,000+ high-quality nodes among them, and intelligently schedule the entire network to achieve high-quality services.
Mature and perfect self-monitoring system
The company has its own 100+ monitoring nodes, and at the same time cooperates with professional network monitoring companies, channels, and customer monitoring systems to grasp the status of CDN nodes, operator backbone networks, customer source sites, HTTPS certificates, DNS resolution, etc. in real time, and collect enough The data information can contribute to a more scientific scheduling decision-making basis.
A truly imperceptible intelligent dispatching system
The intelligent scheduling system is linked with the monitoring and analysis system to make corresponding intelligent scheduling decisions according to the fault level. For availability faults, node faults, abnormal status codes, etc., it will automatically switch according to the set threshold; for abnormal situations such as network jitter, download speed, and response time Do manual optimization. Due to the large number of channels, the Fire Umbrella Cloud Intelligent Fusion CDN can implement a quality-first scheduling strategy after it finds that the fault is difficult to solve, and automatically schedules first, and then solves the problem, so that the user is almost unaware.
The comprehensive management and control platform grasps the real-time operation situation in real time
The Fire Umbrella Cloud integrated CDN platform can control the operation status of the entire integrated CDN platform, and realize all aspects of management and control from basic data to operation and maintenance, operation, whether it is resources, services, status, task system, API system, or monitoring and scheduling , billing and problem follow-up, so that the overall situation is controllable. For customers, Fire Umbrella Cloud maximizes the ease of use. Whether it is adding or removing channels or changing configurations, they only need to communicate with the Fire Umbrella Cloud platform. Once purchased, many are available.
