SaaS has become a typical cloud computing delivery model. In sparse table, one of the most popular multi-tenant data storage schemas for SaaS, all tenants' data are stored into sparse table and mapped to tenant's logical view by metadata. During the data storage schema upgrade in SaaS, all tenants' data need to be migrated into the new data schema before it becomes effective to ensure the integrity of the tenants' data. However, the migration is complex and brings overhead workload. Worse still, it may cause the system unusable. In this paper, we propose metadata evolution technology. We can realize the mapping from the old data schema to new data schema smoothly via metadata evolution during schema upgrade. So the schema upgrade can be completed without data migration.
展开▼
