Placement of SaaS cloud data and dynamically access scheduling strategy

For SaaS applications which share tables in storage, how to effectively manage multi-tenant cloud data has been one of the crucial issues. In SaaS delivery platform multi-tenant data migration strategy, put whole data of every tenant including his/her business data and metadata into independent partition, where DBMS can guarantee ACID. Unfortunately, placing the partitions and the replica of them on randomly nodes maybe easily influenced load balancing of nodes. What's more, it may increase the data access time and cost of updating replication. Therefore, we propose multi-tenant cloud data and replica placement strategy based on the graph, we take network distance and load balancing of nodes into account. We also propose a dynamic scheduling method according to the load state of nodes. We use the Paxos algorithm to further ensure that the SaaS application system availability and consistency. Experiments show that, the proposed strategy can not only improve load balancing, but also can reduce the time for updating replica of partitions, reduce the cost of accessing data.