Concut i —
92 concurren & Secu
icv
ansaction
jon is defined as a logical unit of work. It consis
‘ consists
vt must be completed together so that the database re
either complete successfully or fail. ileal
lore operations
og transact
in a consistent
jabase th
0 teeation MUS mph
ie 4 a the information of stock is stored in the database.
SST tan quantity. When an items sold toa costomen ttieddel Goan
ie Xe operation on the database, However, the quantity ofthe sod ted tothe sold
is Mhrated from the items currently in stock. If both steps are not completed to mua
syste be in an inconsistent state ‘The quantity of items in stock will be wrong if in sre
Pe eubtracted from items in stock. The quantity will be wrong if the teme ns
wn items in stock but not sold. Both of these operations together make up a
both must complete successfully or both must fail. “pa
sted fr0
sale ansaction and
gid Commit and Rollback
«tansaction is committed if it completes successfully and changes the data fit fails
inj eaves the data unchanged i is said that the transaction has been rolled back. It means
2 lack s used to undo the work done in the current transaction.
91.2 Transaction Properties (ACID Properties)
+s transaction must have four properties called ACID properties. These are as follows:
1, Atomicity
‘A transaction must be an atomic unit
orcompletely fail. If any statement in transaction fails,
When a transaction is executed successfully itis said to be committed. In case of failure
ofastatement, all previous successful statements in transaction are rolled back or reverted to
theprevious state of consistency. An incomplete transaction does not take place.
2,Consistency
A transaction must leave the data in a consis!
2 bank database, money should never be "creates
deposit or withdrawal.
of work. A transaction must completely succeed
the entire transaction fails completely.
tent state after completion. For example, in
.4” or “deleted” without an appropriate
3.Isolation
All transactions that modi!
the same data at the same time. Transactions must
‘tansactions. A’ modifying transaction can access
“ansaction is completed.
4. Durability
fy the data are isolated from each other. They do not access
have no dependence or effect on other
the data only before or after another
ion are permanent and
| The durability means that the modifications made by a transact a
id be guaranteed to be complete:
stent Ifthe system is crashed or rebooted, data shoul
'n the computer restarts.
52 Concurrency
Cor : _
same currency is a situation in which
Sitatong mee 18 @ multi-user environment,
the concurrent access may arise to some ser
~ eS
i data at
two or more users access the same piece i tat
concurrency occurs Very commonly:
jous problems. he same dat
ons update I ac
w and then update the |
e of other transactions. 7,
It results in loss of data
ct the same FO)
js unawar
transactions.
attributes Product Code,
Retrieve Qty.
Retrieve QV: Qty. = 100
ty = 100. Update Qty = Qty - 90
Update Qty = Qty + 30
Table: Concurrency Problem
Result; Team B's update is lst at 10:38 which is overwritten by Team A's update
2. Uncommitted Dependency Problem
This problem arises when two or more transactions work on the same table. Oe
transaction retrieves or updates certain part of data before other transaction rollbacks upiit
on the same data. The uncommitted dependency occurs when a second transaction seleds!
row that is being updated by another transaction. The second transaction is reading data ti!
hhas not been committed yet and may be changed by the transaction that is updating ther"
Example :
Suppose the stock table contains a fi i
seuerefstonsmay resulting | aNNY ith value 100, The fle
database with «
rechanism wi"
Team A
7
Row level (Smallest lock granularity)
o Page level
o Table level
© Database level (Largest lock granularity)
‘A lock with large granularity is easy to maintain for DBMS but may cause problems:
lock with small granularity is difficult to maintain for DBMS but causes less problems.
9.4 Serializable Transaction Schedules
d/write operations according, to ce
t, we can examine which of be
correctnes#
A transaction consists of one or more rea
sequence. When two or more transactions are present
operations can be used without harming the original sequence. It ensures the
increases efficiency.
9.4.1 Serial Schedule
Serial schedule is a plan to execute transactions serially. The inte
each transaction is preserved. A serial schedule ensures that each transacti
is the only one accessing the database at one time.
9.4.2 Serializable Schedule
int rmaleable Schedule is a non-serial schedule in whicl
erleaved while ensuring, consistency.
9.4.3 Properties of Serializable Schedule
* Aseriali .
oot lenient schedule is equivalent to some serial schedule.
* Itdoesn read and write conflicts which can effect database consisten
ot suffer from issues such as, lost update, dirty read ete-
nal sequen
ion execut™ a
. tio
hh transaction oP