Java – some queries about serializable transactions?

2020-08-02 • Java

I've read that commit isolation can provide more concurrency than serializable isolation levels

My question is, how does read committed isolation provide more concurrency than serialization? A short scenario would be a great help

Question: – say that two transactions tran1 and tran2 start at time T1 (when I say that the transaction starts, it means that the transaction starts when they trigger some queries, right?) If tran1 updates row1 at time T2 and later updates tran2 at time T3 to get the same row, it will not complete the updated row through tran1 Right？ I say this because tran1 and tran2 will be the result of the current data state at T1 Is that right? Does the transaction take a snapshot of the full database at the beginning?

I have read that encoding serializable transactions requires additional work by the application developer to check for "unable to serialize access" errors and undo and retry the transaction Uncertain when developers will receive an "unable to serialize access" error? Will we get the same error in the following scenario?

Suppose two transactions tran1 and tran2 start at time T1 If tran1 updates and submits row1 at time T2, tran2 updates and submits the same row1 at time T3 In this case, will it throw the "unable to serialize access" error? If so, does Oracle maintain the version internally to prevent serializable transactions so that it knows that the user has updated the row?

Solution

Refer to the Oracle documentation on isolation levels and data concurrency

The content of this article comes from the network collection of netizens. It is used as a learning reference. The copyright belongs to the original author.

THE END

Java