事务

State

Schedule(调度) – a sequences of instructions that specify the chronological order in which instructions of concurrent transactions are executed
Serializability- Each transaction preserves database consistency
- conflict serializability
  - Read-Write conflict
  - Write-Read conflict
  - Write-Write conflict
- view serializability
conflict equivalent(过程一致)- If a schedule S can be transformed into a schedule S’ by a series of swaps of non-conflicting instructions, we say that S and S’ are conflict equivalent
view equivalent (READ一致):
- same transaction to initial READ(Q)
- same transaction to ifinal WRITE(Q)
- must obey Tj WRITE(Q) -> Ti READ(Q)
conflict/view serializable - We say that a schedule S is conflict serializable if it is conflict equivalent to a serial schedule
Every view serializable schedule that is not conflict serializable has blind writes

:
If precedence graph is acyclic, it is conflict serializable , the serializability order can be obtained by a topological sorting of the graph
- use Cycle-detection algorithms

GATE & ESE - Precedence graph method for testing conflict serializability Offered by Unacademy

The precedence graph test for conflict serializability cannot be used directly to test for view serializability.
- checking if a schedule is view serializable falls in the class of NP-complete problems

Recoverable schedule: if haveTi WRITE(Q) -> Tj READ(Q) must have Ti commit - >Tj commit , commit operation of Ti appears before the commit operation of Tj
the following schedule is not recoverable ,is if T8 fails ,it need to abort T9 for atomicity

Cascading rollback – a single transaction failure leads to a series of transaction rollbacks.
Cascadeless schedules — cascading rollbacks cannot occur ,Tj reads a data item previously written by Ti , the commit operation of Ti appears before the read operation of Tj, Ti commit -> Tj READ(Q)
- Every cascadeless schedule is also recoverable

either conflict/view serializable
recoverable and preferably cascadeless
Different concurrency control protocols provide different tradeoffs between the amount of concurrency they allow and the amount of overhead that they incur.

Serializable — default
Repeatable read — only committed records to be read.
- However, a transaction may not be serializable – it may find some records inserted by a transaction but not find others.
- phantom read
Read committed — only committed records can be read.
- Successive reads of record may return different (but committed) values.
- Unrepeatable Read
Read uncommitted — even uncommitted records may be read.
- dirty read