Abstract
Multi-tenant data management is an important part of supporting efficient operation of software as a service application. Multi-tenant data use shared schema to reduce resource usage cost. However, massive data of different tenants are stored in the same schema, which causes useless data of other tenants to be read when a tenant just need access its own disk data. In this paper we focus on disk storage method of multi-tenant data based on shared schema to address the above low efficiency of data access. According to isolation requirement of multitenant data, we store a tenant’s data in some contiguous disk blocks. The experimental results illustrate that query efficiency in range query and join query is 1.5–2 times the existing storage method, and no indexing query efficiency improves 10–70 times.