Method and system for processing replicated transactions parallel in secondary server
AssigneeSolid Information Technology Oy
InventorsJarmo Ruuth, Jarmo Parkkinen, Petri Soini, Antoni Wolski
Life of this patent
Estimating the term…
Abstract
This invention relates generally to database servers and computer systems and, more particularly, describes the mechanism to run transaction operations originating from a primary server used to replicate data in parallel in a secondary server. Especially the invention relates to running concurrent or parallel operations in a secondary server for redundancy, recovery and propagated transactions. This invention describes how parallel operation in a secondary server improves performance and availability and how it maintains transaction order and output congruent with the primary server where transaction operations are originated.
Claims30
A method for processing on at least one secondary server a plurality of database transactions originating from at least one primary server in a database system, comprising steps, in which
- a first timestamp is attached to each transaction of the plurality of transactions in the primary server in the beginning of the transaction, and
- a second timestamp is attached to each transaction of the plurality of transactions in the primary server at the end of the transaction, wherein
- said second timestamp is a commit level of an ending transaction and it is determined to be the highest AND latest commit level among the plurality of transactions incremented by a non-zero number,
- first and second timestamps of each transaction of a plurality of transactions originating from at least one primary server are evaluated according to specific rules, and
- in pursuance of the evaluation according to the specific rules the plurality of transactions originating from at least one primary server are run parallel and/or serially in at least one secondary server.
A database system, comprising at least one primary server and at least one secondary server, wherein at least one secondary server is arranged to evaluate a first and second timestamp attached to each transaction of the plurality of transactions in the primary server, said second timestamp is a commit level of an ending transaction and it is determined to be the highest AND latest commit level among the plurality of transactions incremented by a non-zero number and is arranged to run according to said first and second timestamp transactions parallel and/or serially on at least one secondary server.
Dependencies are inferred from the claim text.