TRANSACTIONS AND LOCKS, TRANSACTION ISOLATION LEVELS

Abstract

The subject of the study is the effectiveness of the use of transaction isolation levels. The purpose of this publication is to determine the feasibility of using common levels of transaction isolation that affect blocking, to prevent data inconsistencies (lost updates, "dirty" reading, non-repetitive reading, phantom reading) in parallel transactions. The results were obtained. Only one transaction isolation level parameter can be set at a time, which remains valid for the current connection until it is explicitly changed. When the isolation level for a transaction changes, the resources read after the change are protected according to the rules of the new level. Resources that are read before the change remain protected according to the rules of the previous level. Database management systems that provide transactivity do not always support all four levels of transaction isolation and may introduce additional levels. Conclusions. Using a high level of isolation (order), you can protect one transaction from the impact of another, but at the expense of significant damage to database performance. At this level, the results of parallel transactions for the database in most cases can be considered to coincide with the sequential execution of the same transactions (alternately in any order). On the other hand, the low level of isolation (reading of unrecorded data) of the transaction creates problems with data inconsistency, while providing higher performance