Load balancing is used to improve the search or other computing performance. Achieving balanced load is a costly processes whatever methods taken, static or dynamic. How to make an acceptable tradeoff in designing is a challenging issue. Most of previous methods focus only on either the application server tier or the storage tier. So the load balancing of integrating the application server tier and storage tier almost is ignored. In this paper, a holistic load balancing strategy is presented based on the object storage system (OSS). This strategy is designed for global peta-scale data-intensive application. And the holistic load balancing method at the object level may be important. At the storage of object-level, both application service and I/O speed are the important consideration with the highest priority in making storage work in high-speed network with huge amount of data to be accessed.