|
Internet era to promote the network computing paradigm shift from parallel computing, distributed computing, grid computing, cloud computing development. Faced with the arrival of cloud computing, virtualization, cloud computing technology seems to be the right-hand man, so that the cloud has a higher flexibility, scalability, low cost, green energy, dynamic deployment and other characteristics. The cloud computing service computing category belong, in practical applications embodied in a web-based service, the purpose is to allow users to pay on demand, while reducing the traditional hardware, software and expertise on investment. Faced with a large number of users to use cloud computing environment, which will be of great significance load balancing, by building a load-balanced cluster to provide high quality services, resources averaged distribution, especially for different hardware performance of the machine, both to maximize the use of resources does not exceed the maximum load resulting downtime. The traditional load balancing method includes the Linux Virtual Server (Linux Virtual Server, referred LVS), the use of load balancing multiple servers to form a virtual server cluster, as well as DNS rotation. LVS to meet the fast growing demand for network access provides an easily scalable, highly available and inexpensive solution. But it is a static load balancing open source software, runs on the Linux operating system, a lot of famous load balancer manufacturer are to be embedded in the hardware which implements static load balancing rotation. In view of this, many domestic institutions have done for dynamic load balancing studies, and its dynamic characteristics is through continuous feedback and remediation servers static weights to achieve, thereby forming a dynamic weight value, and with the performance of the server changes, effective balanced server cluster. The emergence of cloud computing, breaking the traditional means of achieving load balancing, load balancing ideas presented in this paper are: load balancing This is cloud computing services, but in the time and in the use of its service cloud computing dynamically scalable features and dynamic Load balancing load characteristics of real-time feedback to achieve a more flexible load balancing system to provide services. As the load on the server is not a standard definition, this paper Weighted server resources, including CPU usage, RAM usage, and server performance parameters including server response time, number of connections, the number of processes such as server load are given a formal definition, and through a large number of experimental data is given reasonable weight combinations to define server load. This article also feedback by adding different times, in different combinations to find the optimal algorithm, which defines the load-depth study. Finally, virtualization technology presents a new build scalable server clustering method extensive application value, and fully explains the cloud computing era network services scalability in practical applications is important.
|