ndein to

to
e e ic i e n t y

ei
cienty
n
neet
ti e v e

wh i c h
w
he
th e
chnique
t
techni ue
o
onn
Stucture
S t r u c t u f e

attibutes
a t t i b u t e s

da
dda Sene

ecods oM
Co Tom the
d a t a bagL
n
to access

ndetia has been don the indeting

indezi
minimLes

the

when a
quey 19 piotesed

the numbe be disks datakasctable

data
iy
a e s sthe books.
Quick locate and we
see
in
what
to
Simila bute.
t 0 ) mdezing attvibute.

attvi
Tndeing in database
tbaje
mdesing
bn
depending
acess
t6
Ue
HoW uick

Sttoetvre:
Colomn
oo

Indeses eated Ustog

Ddta

Sea ch Kepenence

Candtdete Keyop te
pimary koU()
Contains
the Copy of
Seach tey that ke Can
So-fted Odeh, So
toble AL 8toredm
vaweJ
he

acces tasily
hoids the addYes
addvess
pointers
hoids the
Cortains
a fetop
efeere
7ata
Of dtsk bloCK

T n d t s i y methods:

pr
(SA )i m a o y M d e x

pense ndey

ponse uex
 pimoby mdex:

table, the
base oe pimmy kKy of
T del ig Waded on
the
a pime tndeng
i s oor 1elation

elat
e Unqpe to each vecotd 1:
p
moke the pepotmance oc
Sotted order,
ce stoved1
psima eys epfidet
Dperation
is gpite Dense tndex
Searchr
Sparse indey
tDo pE
p i m e y mdex classigi ed into

Dense indek:
TeCovd5
as n0Of
table is Same
in nde
Numbex oe e o d s
m a i n table
,
mote Space
index eeds'
Dense
)
matn table
nder tasle

ag1a
UP
2t09
USA USAChC9
NepaKathmd 803
Nepa PoRCabid 000

UK

Spavse tndes the

t0 each aetovd in the
matin table,
n s t e a d o p p o i n t i n

able tna
mayr table ina gaP
fecotds in the
td the
nde poirks the btock
bloCKS, and
divided into
table data 13 imdex tabe
he Mat- efevence Op
11 dlota,
ie stoved
gddess

30

46
 so
Aaby ndeK: S1te of
Mappmg
the
AS he S1e oe 4he
table ow
beco
gHOs n 'spavse idevg a d d r e 55
tselp tom
the
fetch HOducd
T ping o s
then

Setondaiy
indEing

Stower o
OveCome
thiS,
mapping
mapp
an
kothe
s
Site of
0 educes the
T
second Indexmg,
intraduced
Cuel og ndeving ts
Smalle
9anges
nto (addres
des
divided moty
Memoy
Eachange
Ch ange s
is
to.
primaty
Me
levels
sto<ed
f i r t

in t h e
Stoed
fetth fattr data
e

actval
and
Jevel
Second
f
pP
Setondee Memo meman)
T a t a block
tn
loo
Poin
Ro poi
Roll
60

priman evel/tnde

(RAM)
320
SLorday lévl 'tnde{H
Chaddig9
p a i n ) ézampe
 tt
Hosh a
hah funcion
h a h fontion
advart oReous techqe wric
oses
arnoud
Hahin s 10
mirimm

etact
ocatien Of a dada e Co«d
the
Pend
Op time
destned doto
d ata on
on disK
disK
Seaach the Jocation oe desined
to
Techiqe
inde 1 8truchne
th out USi
hose
whose
addaeS 18
doda blocKS
blocks
stoted Rn fom o Memory locot On
D a t a îs the memoryocot
appiga hah functicm in

9enexatd
Qeneattd by Knoun as a dadabloc c dot

these
eCotds axe Stoed
whee

ocKet
Memo
oCobns n ee
where data
DatabuctetS 0e
Data bucket; ftoda4
toat
KoowO as t o
Stoed aso
Cene cods) e
4ample:
Division method:
K m)
(mod
HC)-
ayh fun cion:
K hash K
M prime N u m b e

whey
whe
Nemainde)
emainde)
the
mndicates

,k{mod
In above fomula
K dfvided by M
68 tuderdtS
wth
w th
68 3tuddtrdtS Unigue Student numbers
Student
numbers
das Unique
Conside
ee fes
E Uus and 3466
O
mn pe

on,

Hee de
d enn
oot etA
e
K, &aJ404
t5
-ofStudtn
numbes

&tudent tota o
than
+1eaten
be
M to p im e mumber)

(02u(med 1) =l
t(102u)-

24uslmod 7i) 34H -

(2uus)
3466
(mcd ) =48
H(3yC6)
 methods:
SAL hashing
Too types o
aeai)
Stadic Hoshina il
w i l l
OJ.ayS
oddress
buctet
eotand data
The
he

the same Studend.id= 10 USirg

fo
addresS
an
e aenerate oddess
bucd addeSS
bucket
Tf the
nesottantt
nesotant
mod3) .

hashing fanctton
aLn be 1
wiLL

hem oved
.DmariC Hashing ane added
and

data buckets
m 1n Which
Mechanig

to e rte
Grede a
helps you
aynamiCaUy

pundion
the hah
TnthiS haghi
values
aone moOf
n thisS
thiS Tde
and 15
+19
33S2
2oad. eyS veys fumtkion
E Key|.+ hash
* hkey)
si
 -EN/
hl
colligio

C o l Si ov)
ha
h ) 8-1
h l ) s . Colicio)
 FanSa ctions

etecotioYL that acce sses and

Ond

AtOnSa ction iS a Un't op p4ogvam (

arious data tems

poss bly updates
Consistent dB
dB
A ansactio) must See a
()Tr (ul
mcoitert
eLecution the dB May be tempoaily
trancactfo)
During
datallase
tansattion Completes Sucesspolty Cis Committed), the
when the
must be oncigtent:
to the database
has made
¢hanges t, has
a tansaction Commits, the
Apter )
System failuves
be there, even te thexe e

pcrlel (
tonsacti0nS arN eLectt in
utiple '9

Two Haim isves to deol

ivawious Soch 9s, hoduae aines & Sstm Cas hey
failune of kinds
ttansacionS-
COndvent eLecwtion op Multiple
 and duabitrty
Atomicity
mentaion of duali t .
dovabilty.
mple atomicitty Ond
and
Supparts atomicity
Cbmponert
manaqmet
Recovey-

Shadow-databose Scheme

one tan lacti6o 919 ackt

acveve a a time.
ata time
aRS rae
that onl
a e n t Consisttnt
points
poirts to the
called db.potte alusas
A ointe«
)
Copy of
the dlb
db-
a l l Updates made on a 8hadod copy oe the data baseand
Pointer is made to poirt to the updated shadow Copuonly gter t h e
 an taction ieaches pattial Comnmrg
to b
oid Consi st ent Cp Foioted
in Case transaetion fails
deleted.
Can be
db-poimth an be used, and the Shado copy

Scheme

Ldb-pointer db-omte

old Copy o g i
me
7
CoPy
datalase
db (to ber
old Cory o deietedd)

database
apter update
a Bepor Update 9
9

ASSUmes disk do not fail:

databases
databases
mepeitient for lose
for
etemely

handle Con Gvent transa etfons

p o e s not
 Acid poptrties
must engive /hb
preserve the integirtu a data the dB syst em
To preserve
neletedy
nepleted
$aetion ase
tyan saetioqL aie pioper
pioper
all opeYations oe the an
AHomicity ethe
'm the datakase o
none ase
Consistenty bf the db
pYeseuey the Consisteny bf the db
tansachon peselves
bnsistenCA: ELecution Of a
eLeute
e ecute Cnarrertly,
onurvertly, each
éach
Athough mutiple
ttansa (tjons may
T8olatiovl:

sactiol mugt be unauware, of othef ConHertiy

an
eLecutima tanSacti onsS

SuCCesely,
SUCCescelly, the dhanges
Completes
Duabi tity: APter a transa.ctim
dotabase peSiSt even e there ne
made to the
made
Ssem oiluaes

(
 ACCout A to )
E THansactlon SORS foM
an&pe
Yead ()

A A 50
3viteln)

4 deadC B)
S B: B+S0

Cwite(s)
daa We have 9 opevcdtions
fov acessing
database
to local
om
S the data item x
Tead) THansfe

bupge
fom oal bupfer of
the data tem xX
Tanpers
wite back to
the
wite(x): eLeated hen
the twan
sa ction that

database
3 & befove
tiansachion
ails agter Step
atomicity eqyianmerd: net epledt
shouid e g u e the Updates should
Step 6*e System
m the db
A Ond B unchanged by the
the Sun 0
Cbnsisteny Sneayiment

the tansadion
ecution o0
transaction i3
n blud slep 3 & 6, aTlothe
solation Aequiment 1nconisiedt
will e an
database , it
acess the partially Updated
alloed Ao
Sum 0g A*B ill be less 4than 't)
db the
tNansackions Serially.
Conbe ensdved tyiiall u byunna
TeolahonS

Other.
that 18 One agter the

transfer D
coRs tokes plae the
equibemert. Once
DUabiiy
Du
to the db by the tansackio9 Mut peisist olespite fi lues
e S
Aotes
 Schedule5
Conoogical
Gider
mstruttions that Specify he
a Seqyente o cieuted
Ciecuted
ransactions ae
ContuNet
nstructio nS
which
all
tansactioos
Congist
must Consist a l
a l
a fot
Schedsle a Set oe
those tansacttions
nstvuci ons 6
istruckions appean inn each
ac
1 which
Must piesewe the o<dee

ndividuel tiansacdhon
a-tansaction that Sutessfilly Comple-te 'ts eeutim kill have
last etemert
Comitinstueti ons o

eveutil
tansaction that falsto Successly Complete ts
S as lagt lement
abot ingttuctión
will hase an

the bala1
Schedule1
and 7 Yansfer lo1 0
B,
S6RS Pom Ato
tnspe
Let T
A t o B.
o M fotlouwed by
T2:
s
schedde în uhich Ti
A Sevial

T
ead(A)

AS A-S0

uitekA)

eadCB)
B: B+SO

wite (B)
eadCA)
temp A +0-

A A-temp
writeCA)

Tead (3)

8: B+emP

wit e(B)

9chedake
schedule3_
p«esete the
valoe o CA +B)
Te ocuWnt Schedole doesnot
Schedue