Allocating isolation levels to transactions

Abstract

Serializability is a key property for executions of OLTP systems; without this, integrity constraints on the data can be violated due to concurrent activity. Serializability can be guaranteed regardless of application logic, by using a serializable concurrency control mechanism such as strict twophase locking (S2PL); however the reduction in concurrency from this is often too great, and so a DBMS offers the DBA the opportunity to use different concurrency control mechanisms for some transactions, if it is safe to do so. However, little theory has existed to decide when it is safe! In this paper, we discuss the problem of taking a collection of transactions, and allocating each to run at an appropriate isolation level (and thus use a particular concurrency control mechanism), while still ensuring that every execution will be conflict serializable. When each transaction can use either S2PL, or snapshot isolation, we characterize exactly the acceptable allocations, and provide a simple graph-based algorithm which determines the weakest acceptable allocation.

Serializable isolation for snapshot databases

“…The main techniques proposed so far Fekete, 2005;Jorwekar et al, 2007) depend on performing a design-time static analysis of the application code, and then modifying the application if necessary in order to avoid the SI anomalies. For example, shows how one can modify an application to introduce conflicts, so that all executions will be serializable even when the application uses SI.…”

Database Replication Using Generalized Snapshot Isolation

“…Condition S1 is an extension to their syntactic condition, and it applies to GSI in centralized and distributed environments. However, their more recent work [12], which is based on analyzing conflict patterns among transactions, identifies necessary and sufficient static conditions for serializability under databases that support snapshot isolation.…”

'Cause I'm strong enough: Reasoning about consistency choices in distributed systems

“…Fekete [22] considered a hybrid consistency model for relational databases where some transactions execute under snapshot isolation [12] and some under serialisability, a form of strong consistency. He proposed conditions determining which transactions in an application need to execute under serialisability for the whole application to be robust, i.e., produce only behaviours that would be obtained by executing all transactions under serialisability.…”