Scheduling in computational grids addresses the allocation of computing jobs to globally distributed compute resources. In a frequently changing resource environment, scheduling decisions have to be made rapidly. Depending on both the job properties and the current state of the resources, those decisions are different. Thus, the performance of grid scheduling systems highly depends on their adaptivity and flexibility in changing environments. Under these conditions, methods from knowledge discovery yielded significant success to augment and substitute conventional grid scheduling techniques. This paper presents a survey on approaches to extract, represent, and utilize knowledge to improve the grid scheduling performance. It aims to give researchers insight into techniques used for knowledge-supported scheduling in large-scale distributed computing environments.