Design And Implementation Of Web Acceleration Solutions Based On Nginx Deployment Environment

Nowadays what the Internet transmits have changed a lot. As the main content of the early Internet is static public text which doesn’t have strong interactivity, the current Internet transmits a large number of dynamic information. Meanwhile, along with the popularity of network, it has been incorporated into people’s daily life, and network traffic is increasing year by year, the scope of business has also been expanded each year, the phenomena of going to the market and berserk shopping in the Internet appear more frequently. The consequent new features of high concurrency, high traffic and large amount of data make Web become a bottleneck of the Internet applications. So it’s significant to research the Web acceleration.This paper focuses on the key technologies of Web acceleration, and accordingly designs a set of Web acceleration architecture. The research of the key technologies of Web acceleration in this paper, always takes reducing server load, lowering network congestion and decreasing access latency as the main line, takes Web acceleration technologies and applications as the main content, conducts an in-depth discussion of the current situation and existing problems of the network structure and the Web server, analyzes in detail the limitations of network physical structure and the performance differences brought by Web technologies and architectures, and conducts high-performance, optimized-architecture research and design at the macro network layer and micro Web server layer.Currently, Web acceleration is confronted with all aspects of problems. First, the existing network has extensive territory and complex structure, China Telecom and China Unicom build their own network, so that the quality of interconnection is poor, and the main operators has few nodes and insufficient bandwidth, the congestion and failure of the main road line always occur. Secondly, the processing capacity of Web server is limited, and will be undoubtedly exhausted when facing a lot of access to server resources. The relational database will become a bottleneck of performance which will crash at first. Again, with increasingly serious network, from the attacks against Web in the past to the attacks against the DNS at present, the attacks which can influence users’accessing speed shave totally spread out.Based on real users’accessing routes, we propose a complete set of Web acceleration architecture for each aspect of our work, the main work and innovations go as follows:1. DNS is the first step of accessing Web requests, the source of Web domain name resolution lies in the authorization of DNS server and the acceleration of authorized DNS. At first we propose the intelligent region resolution, which based on the IP library, for the sources of accessing users and select the nearest CDN node. Secondly we balance the VIP load of the nearest node by using CNAME resolving methods and ensure that the node LVS does not have the performance bottleneck at a single point. Again we use grouped load on the DNS server, improve the speed of DNS resolution. At the same time we strengthen the protection capability against DNS attacks.2. CDN acceleration is also a widely used acceleration method. This paper takes the high-performance Nginx as proxy server, and takes the Varnish cluster as a cache server. In addition to the traditional static and dynamic separation technology, the NAT back to the source model is given. This model takes the expense of deploying local intranet or two-line nodes to get quick conversion of the carrier line, so that we just need a single line engine room to achieve a multiline accessing quality no matter how many Web main stations we have. Meanwhile we give the real-time detection for the CDN accessing quality, adjust user access to the optimal accessing CDN node, and accomplish the smart link switch when link congestion and failure happen.3. The performance of Web main station influences directly the connection speed of the macro network and the micro-network. Through the architecture analysis given in the literature[1], we point out its advantages and disadvantages, and propose the load based, cache focused, high-performance Nginx as the Web server, distributed master-slave database of high-performance architecture. And then we conduct the experiments of the given architecture and analyze the results. The experiment proves that, this paper gets better results than the traditional model, in the aspects of the number of concurrent in the Website, the response speed, the server resource consumption.Current Web acceleration method is still faced with various problems. Especially with the particularity of Chinese network, we can’t completely change the slow characteristics in the physics. Meanwhile the violent attacks of the network have increasingly normalized which also increase the difficulty of accelerating the network. We hope that through the optimization of architecture and performance, we can keep up with the pace of network progress.