Research And Implementation Of Linux High Availability Cluster

With the rapid development of science and technology, Internet technology has got great development, and people’s requirement of network service increases gradually. As to large network companies, enterprises, groups, the servers must be24hours uninterrupted provide service for users. All of these require the servers to provide high performance, uninterrupted service, so whether the servers are efficient and stable and high availability has been paid more and more attention. It also becomes the key to enhance the competitiveness of enterprises. High performance, uninterrupted and efficientapplication service make the enterprises and social public institutions emphasize handling business, performance of servers, stability of servers, and the system’s high availability requirements to an unprecedented degree. The development and popularization of all these applications make people pay more and more attention on high efficiency and stability of service system. The requirements of customers makes enterprises, organizations and groups in urgent demand for high availability system. If there is any downtime means serious losses, the decline in customers’experience, and it may cause the loss of customers, the decline of entire enterprise reputation, and the loss of the core competence.Current solutions are strengthening server hardware performance and changing server system architecture. High performance servers can provide service reliability and load capacity. But the price is expensive, and reliability cannot meet the requirement of actual production. Another approach is to construct a cluster. There are diverse cluster solutions, different solutions apply to different cluster. Many open source modules and flexible configuration give us great space to improve the cluster.High availability cluster aims to improve the unremittance of cluster’s service, load balancing cluster aims to improve the service performance of the cluster, and the combination of the two can guarantee the stability and high performance of cluster.Based on the needs of our project, we intergrated high-availability cluster and load balancing cluster and designed the structure of the cluster, which has been improved to meet the needs of each functional module. The work of this paper are as follows:1. According to the merits of various clustering solutions, we designed a new cluster structure, which intergrated high-availability cluster and load balancing cluster. And then we analyzed the reasonable storage structure in cluster.2. We modified the open source project heartbeat, which had been used in high-availability cluster. We stripped the resource management function from the heartbeat.3. We added the-management and monitoring module in the dual system, and we could monitor and manage the system on control page.4. We came up with a dynamic load balancing algorithm, which was improvement of dynamic scheduling algorithm. The director will allocation. tasks according to the operating parameters and network conditions of each node in cluster.At last, we conducted the system and made a switch and pressure test.