| The system of HiService is launched by Beijing Hiservice company to provide data management for small and medium enterprises. It is based on SaaS architecture is used in this system to provide data management service for multi-tenant.With the development of HiService system, and the increasement of the number of tenants, the importance of database is becoming more and more significant. Once the database server fails because of man-made reasons or natural disasters, the company and its clients will suffer from irreparable loss. To improve the disaster recovery capability of HiService system, it is necessary to provide a backup for the database. Data migration, that is, migrating data from one database to another, is the key technology for backup system.When SaaS platform owns thousands of tenants, one single database can not meet the storage demand of all tenants. Thus distributed database system will be certainly used. In the distributed database system, with inserting or deleting operation, the distributed system will come to data imbalance. To rebalance database load, we need to migrate data from the heavy load node to the light node. Data migration is a very important technology in the distributed system. This paper mainly talks about data migration in the backup database system and the distributed system. In the backup database system, we provide a backup scheme for one single database. Taking the multi-tenant feature of SaaS platform into account, we designed and impleted data migration system that migrates each tenant’s data seperately. And we made experiment to prove that this system works as we designed. In the distributed database, we first studied previous research, and analysed the disadvantages in their data migration model. And then we put forward a data migration model based on network capacity constraints. We made experiment by comparing our model and previous models, and proved that distributed system operating data migration with our model can reach better performance. |
PDF Full Text Request