Service state consistency maintenance method based on transactions

Abstract

The invention discloses a service state consistency maintenance method based on transactions. The service state consistency maintenance method supports restoration attribute variety by introducing and distinguishing four restoration attributes, dynamically confirming the attribute situations of service examples in the dispatched composite service example set and any other relevant service examples in two sets according to restoration attributes, timestamp size and dynamically-detected conflict relation by establishing a rollback composite service example set and a non-rollback composite service example set with a waiting queue, so that service concurrence, destruction, recovery and submitting in a composite service engine are finished. The service state consistency maintenance method supports the service restoration attribute variety and various control modes, can ensure the serializability and restorability of concurrence dispatch, avoid a starvation phenomenon, permit concurrence dispatch and execution of multiple composite service examples executing irrecoverable transactions and support dynamic conflict detection of the service examples and can effectively reduce linkage destruction of long-term running transactions.

Description

technical field [0001] The invention relates to a method for maintaining state consistency of services running in an operating system, in particular to a method for maintaining service state consistency based on transactions. Background technique [0002] Transaction processing technology originated from the database management system (DBMS: Data Base Management System). A transaction is a collection of operations that constitute a logical unit of work. Its correct execution needs to ensure ACID (Atomicity, Consistency, Isolation, Duration) characteristics, namely atomicity, consistency, isolation, and durability. Atomicity means that all actions in a transaction are executed or none are executed; consistency means that the database must be in a consistent state at any time; isolation means that the internal state of uncommitted transactions is invisible to other transactions; Persistence means that the modification of the database by the committed transaction is permanent....

AI Technical Summary

Problems solved by technology

But there are also shortcomings: in order to prevent non-serializable scheduling, the process lock mechanism often only allows one composite service instance to execute non-compensable services, which greatly limits the concurrency of scheduling; scheduling based on mixed granularity conflict detection Chained aborts cannot be avoided, that is, once a composite service instance is aborted, the affected instances will also be aborted. Chained aborts will cause multiple composite service instances to die and exit. It needs to consume extra resources to compensate, and for the combined service that has been running for a long time, the cost is often very high; the conflict relationship between transactions in the QoS-based concurrent scheduling is statically set before the global scheduling, and for black As far as services with box features are concerned, in actual use, it is difficult and limited to statically set the conflict relationship

Images