The invention relates to the field of databases, and particularly provides a main and standby method between database clusters, which comprises the following steps: S1, after a transaction is started, obtaining a unique transaction ID and returning the transaction ID to a client; s2, a user sends an SQL (Structured Query Language) sql1 to a connection node, the connection node forwards the received sql1 to a corresponding coordinator node for execution, responds to the connection node after execution is completed and returns the sql1 to the client, and meanwhile, an SQL statement or a dependency relationship between Key-Value data corresponding to the SQL statement and transactions is sent to a Merge module of a standby site; s3, the user sends a second SQL sql1, and the process is the same as that of the step S2; s4, the user sends commit/abort statements, and the transaction enters a submission stage; and S5, after receiving the commit, the Merge module of the standby site verifies whether the transaction is complete or not through the key span of the transaction record and the final version number of each range. Compared with the prior art, when the main database breaks down, the main database can be rapidly switched to the standby database, and the disaster recovery capacity of the database is improved.