Data seeding optimization for database replication

摘要

Streaming database replication is provided by merging a stream of user transactions with a stream of copy transactions comprising copy data into a combined stream on a source. A target receives a single stream that includes copy transaction and concurrent user transactions in an order that enables conflicts between data being copied and user transactions to be handled correctly. Furthermore, locks applied to data subject to a copy transaction or user transaction can be released once the copy transaction or user transaction is added to the combined stream.

主权项

1. A method of configuring a primary server, storing a primary replica of a database, to seed a target server storing a target replica of the database, comprising:
transmitting, to the target server, a sequential stream of data commits for the target replica; selecting respective batches of records of the primary replica; for respective batches of records of the primary replica:
acquiring a batch lock over a portion of the primary replica that comprises the batch;inserting the batch into the sequential stream for committing to the target replica; andon condition of completing the inserting, releasing the batch lock; and responsive to receiving a transaction that modifies the database:
acquiring a transaction lock over a portion of the primary replica that is modified by the transaction;determining a sequential position of the transaction in the sequential stream, relative to a batch inserted into the sequential stream that is modified by the transaction, such that applying the data commits of the sequential stream in sequential order makes the target replica consistent with the primary replica;inserting the transaction into the sequential stream at the sequential position; andon condition of completing the inserting, releasing the transaction lock.