Design And Implementation Of Multi-Level Load Balancing Scheduling Strategy In Cloud Environment

In the era of big data and cloud computing today, resources and data are very important and valuable, and these resources and data will eventually be stored onto tens of thousands of servers. These servers will be classified combined to form clusters together into a whole system to provide services. In a cloud environment, volume and size of such clusters are large. the resources and data need to be split more finesse to evaluate. Thus, in a cloud environment, load balancing technology has become more important because through the effective use of it, we can maximize the use of various resources under the cloud environment.Most traditional load balancing techniques are most for small-scale cluster environment or web cluster environment, their capacity for the load balancing of complex applications environments is poor, we use the number of request connections, traffic or response time as a load basis to judge for most of these traditional systems and most of them can only handle static load balancing, for simple web service request type can be more simple and efficient equalization processing, but for more complex request under the cloud environments will not be very good for load balancing and evaluation.In this thesis, how to deal with the load evaluation and load balancing within the cluster for cloud environments is the main research content. It will be proposed a new load balancing scheduling strategy based on a multi-level model: multi-level load balancing scheduling strategy in cloud computing. and also a more effective load evaluation method from a practical point of view. multi-level model build its own architecture of a three-two field, scheduling and balancing strategy will be different according to the corresponding level, each level can receive the request tasks independently, at the same time, load evaluation method will be combined in order to achieve a more balanced distribution and to chose a specific node to carry out for the request, so to avoid excessive concentration on some of the nodes to deal with the request. In addition, the system will be a huge cluster of servers for a reasonable division through the establishment of different domains, so the load balancing strategy can be carried out at the same time in different levels of multiple domains, in order to reduce the performance of the pressure due to the expansion of the cluster and to achieve the ability to maximize the use of the cluster nodes of all the resources in a cloud environment. For reliability, master nodes in the architecture are combined together in a cluster to prevent single point failure. Besides, in this thesis, When all master nodes are invalid, the meta nodes in each sub domain network can still serve as a whole to provide the service and carry out the effective load balancing by the realization of load balancing and scheduling strategy of this system, to ensure the high reliability of the requirements in the cloud computing environment. Finally, a load balancing system which is more suitable for cloud environment is realized.