An index model for multitenant data storage in SaaS

Abstract

For SaaS applications which multiple tenants share sparse tables, fields of varying data types are stored in one flex column based on customization. Thus it is not practical to create native physical indices for tenants. In this paper, we put forward an indexing model for sparse table storage model based on pivot tables in multi-node environments. The core idea is copying field data to be indexed to another table and creating physical indices in that table. This model supports customization and isolation characteristics of multitenant applications, and provides load balance means in index subsystem, and controls the number of index data tables thus memory consuming reasonably. © 2013 Springer-Verlag Berlin Heidelberg.