DBMS CIIT Ch8
DBMS CIIT Ch8
1
Outline
• Introduction to transaction processing
• Transaction and system concepts
• Desirable properties of transactions
• Schedules and recoverability
• Schedules and Serializability
• Transaction support in SQL
• Summary
2
Introduction to Transaction Processing
• Single-user VS multi-user systems
• A DBMS is single-user if at most one user can use the
system at a time
• A DBMS is multi-user if many users can use the system
concurrently
• Problem
How to make the simultaneous interactions of multiple
users with the database safe, consistent, correct, and
efficient?
3
Introduction to Transaction Processing
• Computing systems
• Single-processor computer system
• Multiprogramming
• Inter-leaved Execution
• Pseudo-parallel processing
• Multi-processor computer system
• Parallel processing
4
Concurrent Transactions
B B B
CPU2
A A
CPU1 A
CPU1
time
t1 t2 t1 t2
Interleaved processing Parallel processing
(Single processor) (Two or more processors)
5
What is a Transaction?
• A transaction T is a logical unit of database processing
that includes one or more database access operations
• Embedded within an application program
• Specified interactively (e.g., via SQL)
• Transaction boundaries:
• Begin/end transaction
• Types of transactions
• Read transaction
• write transaction
• Read-set of T: all data items that transaction T reads
• Write-set of T: all data items that transaction T writes
6
A Transaction: An Informal Example
7
Database Read and Write Operations
• A database is represented as a collection of named data items
• Read-item (X)
1. Find the address of the disk block that contains item X
2. Copy the disk block into a buffer in main memory
3. Copy the item X from the buffer to the program variable named X
• Write-item (X)
1. Find the address of the disk block that contains item X.
2. Copy that disk block into a buffer in main memory
3. Copy item X from the program variable named X into its correct
location in the buffer.
4. Store the updated block from the buffer back to disk (either
immediately or at some later point in time).
8
A Transaction: A Formal Example
T1
t0 read_item(X);
read_item(Y);
X:=X - 400000;
Y:=Y + 400000;
write _item(X);
tk write_item(Y);
9
Introduction to Transaction Processing (Cont.)
• Why concurrency control is needed?
• Three problems are
1. The lost update problem
2. The temporary update (dirty read) problem
3. Incorrect summary problem
10
Lost Update Problem
time T1 T2
read_item(X);
X:=X - N;
read_item(X);
X:=X+M;
write_item(X);
read_item(Y);
write_item(X);
Y:=Y + N;
write_item(Y);
11
Temporary Update (Dirty Read)
time T1 T2
read_item(X);
X:=X - N;
write_item(X);
read_item(X);
X:=X+M;
write_item(X);
read_item(Y);
T1 fails and aborts
12
Incorrect Summary Problem
time T1 T2
sum:=0;
read_item(X); read_item(A);
X:=X-N; sum:=sum+A;
write_item(X);
read_item(X);
sum:=sum+X;
read_item(Y); read_item(Y);
Y=Y+N
sum:=sum+Y
Write_item(Y)
13
What Can Go Wrong?
• System may crash before data is written back to disk
= Problem of atomicity
• Some transaction is modifying shared data while
another transaction is ongoing (or vice versa)
= Problem of serialization and isolation
• System may not be able to obtain one or more of the
data items
• System may not be able to write one or more of the
data items
= Problems of atomicity
• DBMS has a Concurrency Control subsytem to assure
database remains in consistent state despite concurrent
execution of transactions
14
Other Problems
• System failures may occur
• Types of failures:
• System crash
• Transaction or system error
• Local errors
• Concurrency control enforcement
• Disk failure
• Physical failures
• DBMS has a Recovery Subsystem to protect
database against system failures
15
Introduction to Transaction Processing (Cont.)
• Why recovery is needed?
1. A computer failure (system crash)
2. A transaction or system error
3. Local errors or exception conditions detected by the
transaction
4. Concurrency control enforcement
5. Disk failure
6. Physical problems and catastrophes
16
Transaction and System Concepts
• Transaction states
• BEGIN_TRANSACTION: marks start of transaction
• READ or WRITE: two possible operations on the data
• END_TRANSACTION: marks the end of the read or
write operations; start checking whether everything
went according to plan
• COMIT_TRANSACTION: signals successful end of
transaction; changes can be “committed” to DB
• Partially committed
• ROLLBACK (or ABORT): signals unsuccessful end of
transaction, changes applied to DB must be undone
17
Transaction States: A state transition diagram
18
The System Log
• Transaction –id
• System log
• Multiple record-type file
• Log is kept on disk
• Periodically backed up
• Log records
1. [start_transaction, T]
2. [write_item, T,X,old_value,new_value]:
3. [read_item, T,X]
4. [commit,T]
5. [abort,T]
6. [checkpoint]
• Commit point of a transaction
19
How is the Log File Used?
• All permanent changes to data are recorded
• Possible to undo changes to data
• After crash, search log backwards until find last
checkpoint
• Know that beyond this point, effects of transaction are
permanently recorded
• Need to either redo or undo everything that
happened since last checkpoint
• Undo: When transaction only partially completed
(before crash)
• Redo: Transaction completed but we are unsure
whether data was written to disk
20
A Sample SQL Transaction
EXEC SQL WHENEVER SQLERROR GOTO UNDO;
EXEC SQL SET TRANSACTION
READ WRITE
DIAGONOSTIC SIZE 5
ISOLATION LEVEL SERIALIZABLE;
EXEC SQL INSERT INTO
EMPLOYEE(FNAME, LNAME, SSN, DNO, SALARY)
VALUES (‘Ali’, ’Al-Fares’, ‘991004321’, 2, 35000)
EXEC SQL UPDATE EMPLOYEE
SET SALARY = SALARY * 1.1 WHERE DNO = 2;
EXEC SQL COMMIT;
GOTO END_T;
UNDO: EXEC SQL ROLLBACK;
END_T: ……;
21
Buffer Replacement Policies
• Least-recently-used (LRU)
• Buffers not used for a long time are less likely to be
accessed
• Rule: Throw out the block that has not been read or
written for the longest time.
• Maintain a table to indicate the last time the block in each
buffer has been accessed.
• Each database access makes an entry in table.
• Expensive ?
Buffer Replacement Policies
• First-In-First-Out (FIFO)
• Rule: Empty buffer that has been occupied the longest
by the same block
• Maintain a table to indicate the time a block is loaded into the
buffer.
• Make an entry in table each time a block is read from disk
• Less maintenance than LRU
• No need to modify table when block is accessed
Buffer Replacement Policies
• Clock Algorithm
• Buffers arranged in a circle
• Each buffer associated with a Flag (0 or 1)
0
• Flag set to 1 when
0 1
• A block is read into a buffer
• Contents of a buffer is accessed 1 0
• A “hand” points to one of the buffers 0 1
• Rotate clockwise to find a buffer with Flag=0 1
• If it passes a buffer with Flag=1, set it to 0
Buffer Replacement Policies
0
0 1
1 0
0 1
1
Desirable Properties of Transactions
• ACID properties
1. Atomicity
A transaction is an atomic unit of processing; it is either
performed in its entirety or not performed at all.
2. Consistency preservation
A transaction is consistency preserving if its complete execution
takes the database from one consistent state to another
3. Isolation
The execution of a transaction should not be interfered with by any
other transactions executing concurrently
4. Durability
The changes applied to the database by a committed transaction must
persist in the database. These changes must not be lost because of any
failure
33
Desirable Properties of Transactions
• Atomicity
• Responsibility of transaction processing and recovery subsystems
of the DBMS
• Consistency
• Preservation of consistency is the responsibility of programmers
• Each transaction is assumed to take database from one consistent
state to another consistent state
• Isolation
• Enforced by the concurrency control subsystem of the DBMS
• Durability
• Responsibility of the recovery subsystems of the DBMS
34
Transaction Processing
• We have discussed that
• Multiple transactions can be executed concurrently by
interleaving their operations
• Schedule
• Ordering of execution of operations from various
transactions T1, T2, … , Tn is called a schedule S
35
Schedules and Recoverability
• Definition of Schedule (or history)
36
Example of a Schedule
• A schedule, S:
r1(X); r2(X); w1(X); r1(Y); w2(X); w1(Y); c1; c2
37
Conflicts
• Two operations conflict if they satisfy ALL three
conditions:
1. they belong to different transactions AND
2. they access the same item AND
3. at least one is a write_item()operation
• Example.:
• S: r1(X); r2(X); w1(X); r1(Y); w2(X); w1(Y);
conflicts
38
Schedules of Transactions
• Complete schedule
A schedule S of n transactions T1, T2, ..., Tn , is said to
be a complete schedule if the following conditions hold:
• The operations in S are exactly those operations in T1, T2, ..., Tn
including a commit or abort operation as the last operation for
each transaction in the schedule.
• For any pair of operations from the same transaction Ti , their
order of appearance in S is the same as their order of appearance
in Ti.
• For any two conflicting operations, one of the two must occur
before the other in the schedule
39
Serializability of Schedules
• Serial Schedule
• Non-serial schedule
• Serializable schedule
• Conflict-serializable schedule
• View-serializable schedule
40
41
Serializability of Schedules (Cont.)
• Serial and Nonserial schedule
A schedule S is serial if, for every transaction T
participating in the schedule, all the operations of T
are executed consecutively in the schedule;
otherwise, the schedule is called nonserial
• Serializable schedule
A schedule S of n transactions is serializable if it is
equivalent to some serial schedule of the same n
transactions
42
Why Do We Interleave Transactions?
Schedule S
T1 T2
read_item(X);
X:=X-N;
write_item(X);
read_item(Y); Could be a long wait
Y:=Y+N;
write_item(Y);
read_item(X):
X:=X+M;
write_item(X);
S is a serial schedule – no interleaving!
43
Serial Schedule
• We consider transactions to be independent, so serial
schedule is correct
• Based on C property in ACID
• Furthermore, it does not matter which transaction is
executed first, as long as every transaction is
executed in its entirety, from beginning to end
• Example
• Assume X=90, Y=90, N=3, M=2, then result of schedule S
is X=89 and Y= 93
• Same result if we start with T2
44
Another Schedule
Schedule S’
T1 T2
read_item(X);
X:=X-N;
read_item(X):
X:=X+M;
write_item(X);
read_item(Y);
write_item(X);
Y:=Y+N;
write_item(Y); S’ is a non-serial schedule
T2 will be done faster but is the result correct?
45
Concurrent Executions
• Serial execution is by far simplest method to execute
transactions
• No extra work ensuring consistency
• Inefficient!
• Reasons for concurrency:
• Increased throughput
• Reduces average response time
• Need concept of correct concurrent execution
• Using same X, Y, N, M values as before, result of S’ is
X=92 and Y=93 (not correct)
46
Yet Another Schedule
Schedule S”
T1 T2
read_item(X);
X:=X-N;
write_item(X);
read_item(X):
X:=X+M;
write_item(X);
read_item(Y);
Y:=Y+N;
S” is a non-serial schedule
write_item(Y);
Produces same result as serial schedule S
47
Serializability
• Assumption: Every serial schedule is correct
• Goal: Find non-serial schedules which are also
correct
• A schedule S of n transactions is serializable if it is
equivalent to some serial schedule of the same n
transactions
• When are two schedules equivalent?
• Option 1: They lead to same result (result equivalent)
• Option 2: The order of any two conflicting
operations is the same (conflict equivalent)
48
Result Equivalent Schedules
• Two schedules are result equivalent if they produce
the same final state of the database
• Problem: May produce same result by accident!
S1 S2
read_item(X); read_item(X);
X:=X+10; X:=X*1.1;
write_item(X); write_item(X);
Schedules S1 and S2 are result equivalent for X=100 but not in general
49
Conflict Equivalent Schedules
50
Conflict Equivalence
Serial Schedule S1
T1 T2
read_item(A);
write_item(A);
write_item(B);
51
Conflict Equivalence
Schedule S1’
T1 T2
read_item(A);
read_item(B);
same order as in S1
write_item(A);
read_item(A):
write_item(A);
same order as in S1
write_item(B);
read_item(B);
write_item(B);
S1 and S1’ are conflict equivalent
(S1’ produces the same result as S1)
52
Conflict Equivalence
Schedule S1’’
T1 T2
read_item(A):
write_item(A);
read_item(A);
different order than in S1
write_item(A);
read_item(B);
write_item(B);
read_item(B); different order than in S1
write_item(B);
Schedule S1’’ is not conflict equivalent to S1
(produces a different result than S1)
53
Conflict Serializable
• Schedule S is conflict serializable if it is conflict
equivalent to some serial schedule S’
• We can reorder the non-conflicting operations to
improve efficiency
• Non-conflicting operations:
• Reads and writes from same transaction
• Reads from different transactions
• Reads and writes from different transactions on
different data items
• Conflicting operations:
• Reads and writes from different transactions on same
data item
54
Example
Schedule A Schedule B
T1 T2 T1 T2
read_item(X); read_item(X);
X:=X-N; X:=X-N;
write_item(X); write_item(X);
read_item(Y); read_item(X);
Y:=Y+N; X:=X+M;
write_item(Y); write_item(X);
read_item(X); read_item(Y);
X:=X+M; Y:=Y+N;
write_item(X); write_item(Y);
B is conflict equivalent to A B is serializable
55
Test for Serializability
• Construct a directed graph, precedence graph, G = (V, E)
• V: set of all transactions participating in schedule
• E: set of edges Ti Tj for which one of the following holds:
• Ti executes a write_item(X) before Tj executes read_item(X)
• Ti executes a read_item(X) before Tj executes write_item(X)
• Ti executes a write_item(X) before Tj executes write_item(X)
56
Sample Schedule S
T1 T2 T3
read_item(Y);
read_item(Z);
read_item(X);
write_item(X);
write_item(Y);
write_item(Z);
read_item(Z);
read_item(Y);
write_item(Y);
read_item(Y);
write_item(Y);
read_item(X);
write_item(X);
57
Precedence Graph for S
X,Y
T1 T2
Y Y,Z
no cycles S is serializable
T3 Equivalent Serial Schedule:
T3 T1 T2
(precedence order)
58
Characterizing Schedules based on
Serializability
• Being serializable is not the same as being serial
• Being serializable implies that the
schedule is a correct schedule.
• It will leave the database in a consistent state.
• The interleaving is appropriate and will result
in a state as if the transactions were serially
executed, yet will achieve efficiency due to
concurrent execution.
59
Characterizing Schedules based on
Serializability
60
Characterizing Schedules based on
Serializability
Practical approach:
• Come up with methods (protocols) to ensure serializability.
• It’s not possible to determine when a schedule begins and when
it ends. Hence, we reduce the problem of checking the whole
schedule to checking only a committed project of the schedule
(i.e. operations from only the committed transactions.)
• Current approach used in most DBMSs:
• Concurrency control techniques
• Examples
• Two-phase locking technique
• Timestamp ordering technique
61
Characterizing Schedules based on
Serializability
• View equivalence: A less restrictive definition of equivalence
of schedules
• View serializability (
• Definition of serializability based on view equivalence.
A schedule is view serializable if it is view equivalent to a
serial schedule.
62
Characterizing Schedules based on
Serializability
Two schedules are said to be view equivalent if the following three conditions
hold:
1. The same set of transactions participates in S and S’, and S and S’ include
the same operations of those transactions.
2. For any operation Ri(X) of Ti in S, if the value of X read by the operation
has been written by an operation Wj(X) of Tj (or if it is the original value
of X before the schedule started), the same condition must hold for the
value of X read by operation Ri(X) of Ti in S’.
3. If the operation Wk(Y) of Tk is the last operation to write item Y in S, then
Wk(Y) of Tk must also be the last operation to write item Y in S’.
63
Characterizing Schedules based on
Serializability
The premise behind view equivalence:
• As long as each read operation of a transaction reads the
result of the same write operation in both schedules, the write
operations of each transaction must produce the same results.
• “The view”: the read operations are said to see the the same
view in both schedules.
64
Characterizing Schedules based on
Serializability
Relationship between view and conflict equivalence:
• The two are same under constrained write assumption which
assumes that if T writes X, it is constrained by the value of X it
read; i.e., new X = f(old X)
• Conflict serializability is stricter than view serializability.
With unconstrained write (or blind write), a schedule that is
view serializable is not necessarily conflict serializable.
• Any conflict serializable schedule is also view serializable, but
not vice versa.
65
Characterizing Schedules based on
Serializability
Relationship between view and conflict equivalence (cont):
Consider the following schedule of three transactions
T1: r1(X), w1(X); T2: w2(X); and T3: w3(X):
Schedule Sa: r1(X); w2(X); w1(X); w3(X); c1; c2; c3;
In Sa, the operations w2(X) and w3(X) are blind writes, since T1 and T3 do
not read the value of X.
66
Transaction Support in SQL
• A single SQL statement is always considered to be
atomic
• There is no explicit Begin_Transaction statement
• SET TRANSACTION statement in SQL2 sets the
characteristics of a transaction
• Access mode
• READ only or READ-WRITE
• Diagnostic area size
• Indicates the number of conditions that can be held
simultaneously in the diagnostic area.
• Isolation level
• READ UNCOMMITTED, READ COMMITTED,
REPEATABLE READ, SERIALIZABLE
67
Type of Violation
Isolation Level Dirty Non- Phantom
READ Repeatable
READ
READ
Yes Yes Yes
UNCOMMITTED
READ
No Yes Yes
COMMITTED
REPEATABLE
No No Yes
READ
SERIALIZABLE No No No
68
A Sample SQL Transaction
EXEC SQL WHENEVER SQLERROR GOTO UNDO;
EXEC SQL SET TRANSACTION
READ WRITE
DIAGONOSTIC SIZE 5
ISOLATION LEVEL SERIALIZABLE;
EXEC SQL INSERT INTO
EMPLOYEE(FNAME, LNAME, SSN, DNO, SALARY)
VALUES (‘Ali’, ’Al-Fares’, ‘991004321’, 2, 35000)
EXEC SQL UPDATE EMPLOYEE
SET SALARY = SALARY * 1.1 WHERE DNO = 2;
EXEC SQL COMMIT;
GOTO END_T;
UNDO: EXEC SQL ROLLBACK;
END_T: ……;
69
Summary
Thank you
70