Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
1
3
,
No.
1
,
Jan
uar
y
201
9
,
pp.
6
5
~
7
1
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
3
.i
1
.pp
6
5
-
7
1
65
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Securin
g SAAS s
erv
i
ce und
er
clo
ud
co
mp
utin
g based
mu
lti
-
ten
ancy sy
stems
JKR S
astry
, M Tri
n
at
h
Ba
su
Depa
rtment
o
f
C
om
pute
r
Scie
n
ce a
nd
Engi
n
ee
rin
g,
KLEF
Univer
sit
y
,
Vadd
eswar
am,
Guntur
Distr
ic
t
,
Indi
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
J
un
1
5
, 2
018
Re
vised Sep
2
0,
2018
Accepte
d Oct
4
, 2
018
Cloud
computin
g
te
chno
logi
es
a
re
bei
ng
used
b
y
m
an
y
who
ne
ed
computing
resourc
es
such
a
s
software
,
pla
tf
orm
and
infra
struct
ure
as
per
th
ei
r
business
req
uire
m
ent
s
in
te
rm
s
of
provisi
oning
and
p
a
y
f
or
the
usage
as
per
a
ct
ua
l
consum
pti
on
of
the
services
base
d
on
the
SLA
s
igne
d
b
y
the
use
r
and
cl
oud
servic
e
provid
er.
Software
running
on
a
phy
si
ca
l
m
ac
hine
is
bei
n
g
provide
d
as
servic
es
to
t
he
end
users.
F
or
the
re
asons
of
cost
ec
onom
i
es
ac
c
ess
to
software
th
at
us
es
a
d
at
ab
ase
is
bei
ng
prov
ide
d
t
o
m
ult
ipl
e
users.
The
acce
ss
to
the
software
is
provide
d
ei
t
her
dire
c
tly
o
r
through
a
virt
u
al
m
ac
hine.
The
softwar
e
be
ing
provide
d
as
servic
e
uses
the
sam
e
dat
ab
ase
f
or
m
an
y
of
the
users
who
h
ave
r
equi
siti
on
e
d
for
the
sam
e
.
As
a
result
,
the
r
e
coul
d
be
enc
roa
chments
b
y
the
users
in
to
t
he
data
of
o
the
rs
.
There
is
a
ne
ed
to
sec
ur
e
the
data
bel
ong
i
ng
to
seve
ral
use
rs
while
al
l
of
th
em
ac
ce
ss
the
da
ta
using
th
e
sam
e
appl
icati
on
.
In
thi
s
pap
er
a
n
eff
icient
m
e
th
od
is
pre
sente
d
f
or
sec
urin
g
the
da
ta
pro
ce
ss
ed
b
y
software
which
is
offe
r
ed
as
a
servi
ce
to
m
ult
iple
users
ei
th
er
dir
ectl
y
o
r
through
v
irt
u
al
m
ac
hine
s.
Ke
yw
or
ds:
Cl
oud
c
om
pu
ti
ng
Mult
i
-
te
nan
cy
SaaS
Sh
a
red data se
r
vices
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed.
Corres
pond
in
g
Aut
h
or
:
J.K
.R
. S
ast
ry
,
Dep
a
rtm
ent o
f C
om
pu
te
r
Scie
nce a
nd E
ng
i
ne
erin
g,
KLEF U
niv
e
rs
it
y,
Vaddes
war
am
, Gun
t
ur D
ist
ric
t
, Ind
ia
.
Em
a
il
:
dr
sastry
@k
l
un
i
ver
sit
y.
in
1.
INTROD
U
CTION
Cl
oud
c
om
pu
ti
ng
pr
ov
i
des
se
ver
al
se
r
vices
to
m
ulti
ple
ten
ants
us
i
ng
th
e
sam
e
ph
ysi
cal
m
achine
thr
ough
im
ple
m
entat
ion
of
c
on
ce
pt
cal
le
d
virtu
al
iz
at
ion.
Pr
ovi
ding
sec
uri
ty
and
pr
i
vac
y
to
data
own
ed
by
sever
al
te
na
nts
is
a
chall
eng
e
wh
e
n
the
res
ources
c
onnect
e
d
to
the
sam
e
m
achine
ha
ve
to
be
sh
a
re
d
am
on
g
sever
al
te
na
nts.
I
n
cl
oud
c
ompu
ti
ng,
the
soft
war
e
m
ark
et
ed
by
the
cl
ou
d
c
om
pu
ti
ng
ser
vi
ce
pro
vid
e
r
is
m
ade
avail
able to
the
u
se
r
th
r
ough i
m
ple
m
entat
ion
of a s
e
r
vice m
od
el
call
e
d
Saa
S.
Ma
ny
us
ers
ar
e
al
lowed
t
o
use
the
sa
m
e
app
li
cat
ion
an
d
there
fore
gi
ve
scop
e
for
enc
roachm
ent
into
a
data
segm
ent
wh
ic
h
is
not
relat
ed
to
so
m
e
us
ers.
Cu
stom
ers
can
no
t
m
on
it
or
or
co
ntr
ol
the
way
the
data
is
go
i
ng
to
be
de
al
t
by
the
se
rv
i
ce
pro
vid
e
r
as
the
us
e
r
has
no
idea
of
t
he
in
f
rastr
uctur
e
bei
ng
us
e
d
t
o
sto
r
e
the
data.
Mult
i
-
te
na
ncy
i
m
plies
t
hat
a
set
of
use
rs
are
al
lo
w
ed
to
util
iz
e
the
sam
e
app
li
cat
ion
hoste
d
by
the
serv
ic
e
pro
vide
r.
T
he
us
e
r
i
s
not
c
on
ce
rn
e
d
with
the
ki
nd
of
un
der
ly
in
g
in
fr
a
struct
ure
bein
g
us
e
d
by
th
e
app
li
cat
io
n
[
1].
SaaS
is
a
ki
nd
of
delive
ry
m
od
el
s
im
ple
m
e
nted
by
cl
oud
com
pu
ti
ng
se
r
vice
pro
vid
e
rs.
Custom
ers
acce
ss
the
ap
plica
ti
on
thr
ough
In
te
r
net.
T
he
so
ft
war
e
a
nd
t
he
relat
ed
data
are
centrall
y
hoste
d
by
the
se
rv
ic
e
pro
vid
er
.
A
vir
tual
m
achine
(
VM)
is
a
ki
nd
of
im
ple
m
entation
of
s
of
twa
r
e.
T
he
VM
ca
n
be
c
onfig
ur
e
d
with
an
O
per
at
in
g
s
yst
e
m
or
any
ot
her
pro
gram
wh
ic
h
act
ually
ru
ns
on
the
Ph
ysi
cal
m
achine
on
w
hich
the
VM
is
pro
vision
e
d.
T
he
s
of
twa
re
t
ha
t
is
config
ur
e
d
on
t
he
VM
us
es
the
al
l
th
e
com
pu
ti
ng
f
aci
li
ti
es
existed
on
t
he
ph
ysi
cal
m
achines t
hat inclu
de
m
e
m
or
y, stora
ge,
netw
ork
etc.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
1
,
Ja
nu
a
ry
201
9
:
6
5
–
7
1
66
A
S
of
t
war
e
cal
le
d
Hy
perviso
r
w
hich
is
instal
le
d
on
t
he
physi
cal
m
achine
on
to
p
of
op
e
rat
ing
syst
e
m
is
respo
ns
ible
for
m
anag
in
g
al
l
the
res
ourc
es
that
are
sh
a
red
by
the
ap
pl
ic
at
ion
s
r
unni
ng
on
dif
fer
e
nt
virtua
l
m
achines.
Ma
ny
resou
rces
s
upporte
d
on
th
e
ph
ysi
cal
m
ac
hin
es
as
s
uch
m
us
t
be
m
anag
ed
s
uc
h
that
on
e
us
er
do
not
co
nf
li
ct
ano
t
her
us
e
r
i
n
relat
io
n
to
t
he
sh
ar
ed
res
ources
al
locat
ed
to
them
.
The
r
eso
ur
ces
t
hat
c
an
be
sh
are
d
am
on
g
the
us
ers
incl
ude
virt
ual
m
ac
hin
es
,
stora
ge,
m
e
m
or
y,
netw
ork
ba
ndwidt
h
et
c.
The
acce
ss
ibil
ity
of
t
he
s
harable
res
ources
am
on
g
seve
ral
te
na
nts
m
us
t
be
c
ontr
olled
th
rou
gh
us
e
of
te
ch
niq
ue
s
s
uch
as
a
ccess
con
t
ro
l,
Vi
rtua
l
stora
ge
c
ontr
oller
a
nd
us
e
of
VLA
NS.
Cl
oud
c
om
pu
ti
ng
syst
e
m
s
al
so
ar
e
sub
j
ect
ed
to
at
ta
cks
wh
ic
h
i
nclu
de
side
c
hannel
a
tt
ack,
brute
f
or
ci
ng
at
ta
ck,
ne
twork
pro
bing
et
c.,
from
whic
h
the
data
an
d
th
e
app
li
cat
io
ns
m
us
t
be
protect
ed.
T
he
m
os
t
i
m
po
rtant
thin
g
is
to
achieve
data
isolat
ion.
Mult
i
-
te
nan
cy
is
an
i
m
po
rtant f
eat
ure of
Saa
S in
cl
oud
com
pu
ti
ng
. Mu
lt
i
-
te
nan
ts
can
sh
are sin
gl
e instance o
f
the sam
e app
li
c
at
ion
there
by
s
har
e
the
sam
e
data
stora
ge
a
rea.
Mult
i
-
te
nan
cy
pro
vid
es
t
he
use
r
the
ease
of
op
e
rati
ons
a
nd
re
du
ces
delivery c
os
t
f
or a
huge n
umber
of tena
nts.
Cl
oud
c
om
pu
ti
ng
sho
uld
sup
port
I
so
la
ti
on
of
te
na
nt
data,
w
orks
pace
(m
e
m
or
y),
Proces
s
exec
utio
n,
Tena
nt
-
awa
re
secur
it
y,
m
on
it
or
in
g,
m
anag
e
m
ent,
re
porting
an
d
sel
f
-
s
erv
ic
e
a
dm
ini
strat
ion
,
Is
olati
on
of
te
nan
t
c
us
tom
iz
at
ion
s
a
nd
e
xt
ens
io
ns
t
o
business
l
og
ic
,
t
enan
t
-
awa
re
ve
rsion
c
ontrol,
Tena
nt
-
a
war
e
er
ror
trackin
g
a
nd
r
ecov
e
ry
et
c.
s
o
as
t
o
e
ns
ure
that
the
data
is
act
ually
prote
ct
ed.
Mult
it
en
ancy
can
be
ac
hieve
d
thr
ough v
ari
ous
m
od
el
s that i
nclu
de
sh
a
red
no
t
hing,
sh
a
re
d
hard
war
e
, s
ha
re
d
O
S,
sh
a
re
d
databa
se, and s
ha
re
d
ever
yt
hi
ng and
cu
st
om
m
ult
itenan
cy
m
od
el
s
.
Data
relat
ed
t
o
m
any
us
ers
cou
l
d
be
sto
re
d
in
t
he
sam
e
database
a
nd
m
anag
ed
thr
ou
gh
t
he
sam
e
app
li
cat
io
n
tha
t
has
bee
n
give
n
acce
s
s
to
m
any
us
e
rs.
S
om
e
m
any
us
e
rs
m
ay
be
gi
ve
n
ac
ces
s
to
t
he
sam
e
ta
bles
existi
ng
in
the
sam
e
database
.
T
he
us
er
s
f
or
the
sam
e
app
li
cat
ion
s
al
s
o
are
giv
e
n
the
op
ti
on
of
config
ur
i
ng
th
e
ap
plica
ti
on
a
s
pe
r
thei
r
bu
si
ness
re
qu
irem
ents.
T
he
m
ai
n
issue
i
n
the
cas
e
of
m
ulti
-
te
nan
cy
is
the
data ris
k,
one u
ser
t
am
per
ing
t
he
data o
f
oth
e
rs.
M
ulti
-
te
nan
cy
is
al
l
about
is
olati
ng
t
he
data
in
su
c
h
a
wa
y
that t
he owne
r of t
he data
on
l
y wil
l hav
e
acc
ess and
kee
ping c
om
plete
co
nfi
de
ntial
it
y.
Data
Ma
na
ge
m
ent
as
su
c
h
will
be
the
ke
y
issue k
ee
ping in
vie
w
of
co
nfi
den
ti
al
it
y
and
p
ri
vacy
of
t
he
data. Ma
ny rules an
d reg
ulati
on
s
m
us
t be in
-
bu
il
t i
nto cl
oud com
pu
ti
ng m
ulti
-
te
nan
cy
env
iro
nm
ent so
th
at
the
needs
an
d
the
regulat
or
y
re
quirem
ents
of
di
ff
ere
nt
us
e
rs
c
an
be
m
et
.
Nev
ert
heless,
it
is
crit
ic
al
that
the
need
to
seg
reg
at
e
t
he
data
an
d
pro
vid
e
d
pro
pe
r
acce
ss
co
nt
ro
ls
so
no
unaut
horized
a
ccess
can
be
gaine
d.
Ma
ny Chall
en
ges
a
re to be
m
et
w
he
n o
ne
im
plem
ents Mult
i
-
Tena
ncy.
En
d
us
e
rs
re
quires
Pe
rfor
m
ance
isolat
io
n,
Av
ai
la
bili
ty
of
al
l
the
resource
s,
Scal
abili
ty
in
te
rm
s
of
te
n
et
s
require
m
ents,
support
for
val
ue
ad
de
d
ap
plica
ti
ons
and
nee
d
f
or
pri
vacy
an
d
secur
it
y
of
t
he
data
acce
ssed
by
a
ppli
cat
ion
s,
abil
it
y
to
custom
ize
the
a
pp
li
cat
ion
s
to
run
t
he
way
their
a
ppli
cat
ion
s
a
re
des
ign
e
d
for
the
cu
sto
m
ers.
So
l
ution
dev
el
op
e
rs
ar
e
c
on
ce
rned
with
the
issu
e
of
acce
ss
c
ontr
ol,
cust
om
i
zabil
it
y
consi
der
i
ng
da
ta
base,
Busi
ne
ss
log
ic
,
us
e
r
interface
,
w
orkf
l
ow
s
,
te
na
nt
pro
vision
i
ng,
an
d
us
age
base
d
m
et
ering
. S
er
vi
ce
pro
vid
er
s’
needs
to d
eal
w
it
h
data
s
har
i
ng, b
ack
up,
a
nd
re
sto
rin
g
te
na
nt
data
,
en
ha
nc
ing
t
he
us
a
ge
of
t
he
ha
rdwar
e
,
re
du
ci
ng
the
operati
onal
c
ost
,
de
velo
pm
ent
of
hum
an
r
eso
ur
ces
,
re
du
ce
the
dev
el
op
m
ent
effor
t,
reduce
th
e
tim
e
to
m
ark
et
,
enab
le
m
ent
of
the
m
utli
-
te
nan
cy
s
upport
to
the
us
e
rs
without
the n
ee
d
t
o
m
ake a
ny code c
ha
ng
e
s.
Mult
i
-
te
nan
c
y
can
be
i
m
ple
m
ented
con
s
iderin
g
virt
ualiz
at
ion
,
sh
a
rin
g
operati
ng
s
yst
e
m
s
and
app
li
cat
io
ns
.
Ma
ny
m
et
ho
ds
hav
e
bee
n
i
m
plem
ented
to
a
chieve
m
ulti
-
ten
ancies
w
hich
include
visu
al
i
zat
ion
,
data
isolat
ion
,
and
m
anag
in
g
databases
.
Virt
ualiz
at
ion
is
creati
ng
m
or
e
log
ic
al
m
achines
that
ru
n
on
a
sing
le
ph
ysi
cal
m
achine
wh
ic
h
is
c
onnecte
d
with
m
or
e
nu
m
ber
of
res
ources
a
nd
al
so
t
hat
m
any
operati
ng
syst
e
m
s
run
on
a
sin
gle
m
achine.
Th
e
resour
ces
c
onnecte
d
to
the
ph
ysi
cal
m
ac
hin
e
are
s
har
e
d
am
on
g
the
virtu
a
l
m
achines.
A
co
nf
i
gure
d
vi
rtual
m
achine
can
opt
to
r
un
a
sp
eci
fic
oper
at
ing
syst
em
.
A
sepa
rate
virt
ual
m
achine
can
be
al
locat
ed
to
eac
h
of
t
he
te
nan
t.
Im
plem
enting
virtu
a
li
zat
ion
requir
e
run
ning
a
se
par
at
e
s
of
t
ware
su
c
h
as
VMwar
e
on
the
physi
cal
m
achine.
Th
e
so
ftwa
re
is
respon
si
ble
f
or
pr
ov
i
ding
serv
ic
es
that
inclu
de
scal
abili
ty
,
flexibili
ty
,
resou
r
ce
sh
a
rin
g.
W
hen
m
aking
a
vaila
ble
the
ac
cess
to
a
data
base,
va
rio
us
aspects
hav
e to
be
sup
ported
that i
ncl
ud
e se
par
at
i
on
of
the se
rv
ic
es
pro
vid
e
d
to d
if
fer
e
nt t
enan
ts,
scal
ing
the acc
ess to
the
database
s
base
d
on
the
nu
m
ber
of
te
na
nts,
co
nfi
rm
i
ng
to
t
he
SL
A
te
rm
s
and
conditi
ons,
s
upport
for
te
nan
ts
c
us
tom
iz
at
ion
li
ke
suppo
rt
to
te
na
nt
def
i
ned
t
rig
ge
rs
an
d
st
or
e
d
proce
dures
.
Eac
h
te
na
nt
to
whom
a
virtu
al
m
achine
i
s
pr
ovide
d,
add
it
io
nal
servi
ces
need
s
to
be
pr
ovi
ded
rela
ti
ng
to
bac
kup
and
ret
rieval
of
data,
enfor
ci
ng
the
ap
plica
ti
on
upgra
des,
secu
r
it
y
enf
orcem
e
nt
an
d
sup
por
t
for
im
ple
m
e
nting
la
w
an
d
act
.
Mult
it
enan
cy
is
al
l
abo
ut
sever
al
te
na
nts
sh
ari
ng
the
sa
m
e
app
li
cat
ion
that
is
dev
el
op
e
d
us
i
ng
da
ta
base
m
anag
em
ent
so
ft
war
e
[2
]
.
The
acce
ssi
bili
ty
of
a
da
ta
base
can
be
pro
vid
e
d
th
ough
m
any
database
m
anag
em
ent
m
echan
ism
s that inclu
de
th
e f
ol
lowing:
a)
Allocat
ion
of
s
epar
at
e
databa
s
es r
esi
den
t
on t
he
sam
e
m
achine.
b)
Alloc
at
ion
of
da
ta
bases on di
f
fer
e
nt m
achines.
c)
A
ll
ocati
on
of
s
egm
ented
data
base
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Securi
ng
SAA
S servi
ce u
nder
cl
oud
c
omp
uting b
as
e
d m
ulti
-
te
nancy
syste
ms
(
JK
R
Sastry
)
67
d)
Allocat
ion
of
Horizo
ntall
y p
arti
ti
on
ed
d
at
a
bases resi
de
nt
on the
sam
e
m
achine.
e)
Allocat
ion
of
Horizo
ntall
y p
arti
ti
on
ed
d
at
a
bases resi
de
nt
on the
dif
fer
e
nt
m
achines.
f)
Allocat
ion
of
s
har
e
d databas
e
w
it
h sepa
rate
schem
a.
g)
Allocat
ion
of
s
har
e
d databas
e
and s
har
e
d sch
e
m
a.
A
database
wi
th
it
s
ass
ociat
e
schem
a
can
be
al
lowe
d
for
acce
ssing
by
s
ever
al
us
e
rs.
T
his
kind
of
sh
ari
ng
is
usu
al
ly
i
m
ple
m
ented
within
an
enter
pr
ise
set
up.
H
oweve
r
th
is
kin
d
of
a
scal
abili
ty
of
a
sing
le
database
le
a
ds
to
unaut
horize
d
acce
ss
t
o
the
data.
T
h
e
gra
nt
per
m
issi
on
ad
op
te
d
by
t
he
use
rs
lim
it
s
the
acce
ss
by
the
us
ers
.
If
acce
ss
per
m
is
sion
s
e
xist,
there
cou
l
d
be
dat
a
encroachm
ents
by
the
us
ers.
The
adm
inist
rators
of
th
e
data
base
will
hav
e
acce
ss
to
the
e
ntire
data
losin
g
th
e
secrecy
of
th
e
sam
e.
More
m
echan
ism
s
hav
e
to
be
a
dap
te
d
s
o
a
s to
e
nsure
the
secrecy
of t
he data
.
Mult
i
-
te
nan
cy
within
SaaS
is
achieved
th
r
ough
use
of
a
database
with
data
isolat
ion
achieve
d
at
app
li
cat
io
n
la
ye
r.
The
ser
vi
ce
prov
i
der
s
s
hall
hav
e
to
bu
il
d
int
o
the
app
li
cat
ion
,
t
he
m
echan
is
m
s
that
i
m
ple
m
ent
data
isolat
ion
c
onsiderin
g
eac
h
of
the
us
er
s
w
ho
hav
e
bee
n
give
n
acce
ss
t
o
th
e
app
li
cat
io
n.
On
e
of
the
isolat
io
n
m
et
ho
d
t
hat
c
an
be
im
ple
m
ented
is
the
da
ta
acce
ss
is
i
m
ple
m
ented
f
or
e
ach
us
e
r
t
hro
ugh
a
separ
at
e i
ns
ta
nc
e of the
sam
e d
at
a access
cl
ass.
The
t
hr
eat
of
data
c
orrupti
on,
data
l
os
s,
da
ta
inco
ns
ist
en
cy
is
ex
pected
to
in
crease
wh
e
n
m
or
e
nu
m
ber
of
us
e
rs
acce
sses
the
sa
m
e
data
wit
h
dif
fer
e
nt
pointers
to
the
da
ta
[2
-
3
]
.
The
s
erv
ic
es
im
ple
m
ented
thr
ough
cl
ou
d
com
pu
ti
ng
inf
rastr
uctu
re
bypass
the
se
cur
it
y
co
ntr
ols
(P
e
rson
nel,
ph
ysi
cal
a
nd
log
ic
al
)
exer
ci
se
d by th
e u
se
rs. This
leads t
o
a
risk w
hen d
at
a c
ontr
ol r
es
ponsi
bili
ty
is left t
o
cl
oud
c
om
pu
ti
ng
s
yst
e
m
.
Much
nu
m
ber
of
iss
ues
arise
s
du
e
to
Mult
i
-
Tena
ncy
due
to
the
reas
on
t
hat
sam
e
har
dware
is
us
e
d
for
al
l
the
use
r
s
w
ho
a
re
giv
e
n
with
the
acc
ess
f
or
th
e
sa
m
e
app
li
cat
ion.
H
ow
e
ve
r
som
e
kin
d
of
se
par
at
io
n
betwee
n
the
us
ers
e
xists at
the
appli
cat
ion
a
nd
Virtual lay
er
s [1].
In
the
case
of
Mult
i
-
Tena
ncy
bo
th
the
Vict
im
and
at
ta
cker
us
es
the
sam
e
app
li
cat
ion
th
at
ru
ns
on
a
sing
le
se
r
ver
.
The
risk
cause
d
by
the
at
ta
c
ker
can
not
be
m
itigated
by
tr
aditi
on
al
m
et
ho
ds
as
t
hese
m
et
hods
cannot
pe
netra
te
into
the
serv
ers
.
The
m
on
it
or
in
g
to
fin
d
at
ta
cking
if
an
y
is
lim
it
ed
to
network
la
ye
r
on
ly
.
Ther
e
are
th
ree
d
if
fer
e
nt
ways the att
acke
r
a
nd the
v
ic
ti
m
ca
n be situat
e
d w
it
hin
the
cl
oud.
In
case
one,
t
he
at
ta
cker
and
t
he
Vict
im
are
si
m
ply
the
internet
us
er
s
w
hich
sim
ply
m
ean
s
tradit
ion
al
secur
it
y
m
e
thods
can
be
us
ed
to
protect
each
ot
her’s
data.
I
n
case
t
wo,
th
e
victim
and
th
e
at
ta
cker
are
i
n
the
sam
e
cl
ou
d
bu
t
on
dif
fer
e
nt
se
rv
e
rs.
The
victi
m
and
the
at
ta
cked
are
physi
cal
ly
separ
at
ed
due
t
o
al
locat
ion
of
diff
e
re
nt
virt
ua
l
m
achines
to
each
on
e
of
th
e
m
.
In
t
his
cas
e
Virtu
al
secu
r
it
y
m
easur
es
a
re
to
be
em
ploy
ed
by
the
ser
vice
pr
ovide
r.
In
a
noth
er
case
the
vic
tim
and
the
at
ta
cker
a
re
on
the
sam
e
cl
ou
d
but
sh
a
re
the
sam
e
serv
e
r w
hich
is
the case
of m
ulti
-
te
nan
cy
.
Secu
rin
g
this
kind
of
sit
uation
is
ha
r
d
as
no
netw
ork
as
su
ch
e
xits
for
com
m
un
ic
at
io
n
to
h
a
ppe
n.
The
tra
ff
ic
as
s
uch
ha
pp
e
ns
w
it
hin
physi
cal
m
achine
only
.
Virtual
netw
ork
sec
ur
it
y
def
e
nces
as
s
uch
c
anno
t
protect
the
data
that
is
attacked
within
the
pur
view
of
a
physi
cal
m
achine.
Ma
ny
issues
are
to
be
addr
esses
wh
e
n
data
of
m
ul
ti
ple
us
ers
is
st
or
e
d
withi
n
th
e
sam
e
reposit
or
y.
Sam
e
crypto
gr
a
phic
al
gorithm
s
are
us
e
d
to
store
t
he
data
in
the
c
omm
o
n
sto
rage.
No
ph
ysi
cal
se
parat
ion
ei
ther
ph
ysi
cal
or
lo
gical
exists.
Als
o
ther
e
exists
a
n
oppo
rtu
nity
to
at
ta
ck
th
e
data
w
he
n
data
is
retrie
ved
f
ro
m
the
d
at
abase
a
nd
de
crypted
to
plai
n
te
xt
for
pr
ocessin
g.
The
processi
ng
job
ca
n
be
interr
up
te
d
to
acce
ss
the
pl
ai
n
te
xt
an
d
t
hu
s
can
be
at
ta
cked.
Ther
e
s
hould
be
c
om
po
sit
e
and
com
plete
pr
i
vacy
an
d
se
cur
it
y
to
t
he
da
ta
of
m
ulti
ple
cl
ie
nts
sto
red
in
th
e
sam
e d
at
abase
bein
g
acce
sse
d by the
sam
e app
li
cat
io
n wh
i
ch
is s
ha
red by
m
any custo
m
e
rs.
2.
RELATE
D
W
ORK
The
m
ai
n
issue
that
m
us
t
be
addresse
d
w
he
n
m
ulti
-
te
nan
cy
is
us
ed
is
to
protect
the
ap
pl
ic
at
ion
an
d
the
ha
rdwa
re
on
w
hich
t
he
app
li
cat
io
n
is
dep
l
oyed
as
th
e
cl
ie
n
ts
are
a
ll
ow
ed
to
s
hare
both
th
e
el
em
ents.
Mult
i
-
te
nan
cy
thu
s
possesses
m
any
chall
en
ges
to
s
ecu
re
and
preser
ve
t
he
pri
vacy
of
the
data
owne
d
by
diff
e
re
nt
us
e
rs
.
Ma
ny
m
et
hods
ha
ve
been
pr
ese
nted
in
the
li
te
ratu
re
t
hat
ai
m
s
at
is
olati
ng
data
st
or
a
ge,
al
locat
ing
se
pa
rate data st
or
a
ge
for
eac
h
te
na
nt etc.
[9
]
.
Each
m
et
ho
d
release
s
a
dif
f
eren
t
sec
uri
ty
issue
al
to
gethe
r
that
i
nvolv
es
us
e
of
diff
e
re
nt
ty
pes
of
encr
y
ption
te
chn
i
qu
e
s.
T
he
s
erv
ic
e
pro
vid
e
r
ca
n
pr
ov
i
de
an
inte
rf
ace
wi
thin
the
ap
plic
at
ion
us
in
g
w
hi
ch
the
us
ers
ca
n
c
onfi
gure
the
a
ppli
cat
ion
f
or
im
posing
s
om
e
sec
ur
it
y
co
ns
trai
nt
s.
The
sec
uri
ty
enfor
cem
ent
can
be
exter
nalise
d
w
it
ho
ut
im
po
sin
g
any
l
oad
on
the
ap
plica
ti
on
Moh
am
ed
Al
m
or
sy
et
al
[1
3].
Cl
oud
c
ompu
ti
ng
arch
it
ect
ure
m
us
t
inclu
de
va
r
iou
s
iss
ues
r
el
at
ed
to
enfo
rce
m
ent
of
the
se
cur
it
y
issues.
The
ve
ry
first
at
temp
t
to inclu
de se
cu
rity
in
to cl
oud com
pu
ti
ng in
f
rastr
uctur
e
w
a
s
att
e
m
pted
by
Kam
ara et al
. [14
]
.
They
ha
ve
c
overe
d
bo
t
h
c
onsu
m
er
an
d
e
nter
pr
ise
sce
na
rios
a
nd
the
y
hav
e
us
e
d
nons
ta
nd
a
rd
encr
y
ption
al
gorithm
s
su
ch
a
s
searcha
ble
e
ncr
y
ption
a
nd
at
tribu
te
enc
ryption.
A
n
al
go
rithm
that
us
e
s
us
er
at
tribu
te
s a
nd t
heir
si
gn
at
ur
e
has bee
n pr
ese
nted by Za
rand
ioon et al
. [
15
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
1
,
Ja
nu
a
ry
201
9
:
6
5
–
7
1
68
The
al
gorithm
is
inclu
ded
int
o
a
prot
oco
l
ca
ll
K2
C
(K
ey
t
o
Cl
ou
d
-
us
e
r
cent
ric
pr
i
vac
y
pr
ese
r
ving
crypto
gr
a
phic
acce
ss
co
ntr
ol
protoc
ol.
T
he
end
us
e
rs
can
secur
el
y
s
har
e,
m
anag
e
an
d
st
or
es
t
heir
da
ta
in
th
e
cl
oud
com
pu
ti
ng
i
nfrastr
uctu
re
w
hich
is
ba
sic
al
ly
un
struc
ture
d.
E
ncr
y
ption
al
gorithm
s
are
quit
e
fr
e
qu
ently
e
m
plo
ye
d
to
se
cur
e
t
he
us
er
da
ta
.
The
e
nc
ryption
al
gorith
m
s
can
be
us
e
d
to
tran
sf
or
m
the
us
e
rs
’
crit
ic
al
dat
a
so
that
t
he
data
can
be
m
ade
to
be
in
-
acce
ssi
ble
to
un
a
utho
r
iz
ed
us
e
rs
e
ve
n
in
t
he
sit
uations
of
a
vaila
bili
ty
of
su
c
h
data
to
unaut
horized
use
rs.
Access
c
ontr
ol
is
one
of
appr
oach
e
s
th
at
can
be
e
nfo
rced
t
o
pre
vent
un
auth
or
iz
e
d
acc
ess
to
data.
A
ccess
co
ntr
ol
e
nfor
ce
d
for
c
ontr
olli
ng
the
a
ccess
w
hile
th
e
us
er
s
are
i
n
m
ul
ti
-
te
nan
cy
m
od
e i
s not a
n
ef
fecti
ve
m
et
ho
ds as
the access c
ont
ro
l i
s m
erely
achieve
d
th
r
ough
us
in
g
IDs
[
12]
.
Data
Iso
la
ti
on
is
the
crit
ic
al
i
ssu
e
that
m
us
t
be
ad
dr
es
sed
wh
e
n
it
com
es
to
m
ulti
-
te
nancy
.
Eff
ect
ive
data
m
anag
em
ent
syst
em
s
m
us
t
be
im
ple
m
ented
t
o
c
ontr
ol
the
acce
ss
t
o
the
data
by
m
ul
ti
ple
us
ers
thr
ough
acce
ss
to
the
sam
e
app
li
cat
ion
.
A
ppr
opr
ia
te
an
d
e
xten
sive
pr
iva
cy
and
sec
ur
it
y
to
the
data
m
us
t
be
i
m
ple
m
ented.
The
public
cl
ouds
as
s
uch
ca
n
be
at
ta
c
ked
t
hro
ugh
se
ver
al
m
eans
as
no
netw
ork
isolat
i
on
is
i
m
ple
m
ented
[
10
]
.
User
s
w
ho
a
re
at
ta
cker
s
ca
n
l
aun
c
h
at
ta
c
ks
on
co
-
re
side
nt
us
ers
as
no
tra
ff
ic
or
ba
ndwi
dth
is
olati
on
is
i
m
ple
m
ente
d
as
the
m
ulti
-
te
nan
cy
is
an
i
ssu
e
that
is
im
plem
ented
within
a
si
ng
le
se
rv
e
r
[
11]
.
DPE
T
(d
at
a
Partit
ion
enc
ry
ption
te
c
hn
i
ques
is
on
e
su
c
h
m
et
ho
d)[
2].
In
this
m
et
ho
d
each
rec
ord
is
encr
y
pted
twic
e
befor
e
storing
the
sa
m
e
in
a
po
rtio
n
that
is
al
loca
te
d
to
the
te
na
nt.
Entire
data
base
is
portio
ne
d
(
us
er
sp
ace
)
and
one
portio
n
is
al
locat
ed
one
te
na
nt
on
ly
.
A
schem
e
is
us
ed
for
portio
ning
a
nd
a
ll
ocati
ng
the
pa
rtit
ion
to
a
s
pe
ci
fic
te
nan
t.
T
he
r
ecord
is
enc
r
ypte
d
usi
ng
a
p
ubli
c
an
d
pri
vate
key
known
to
both
the
te
nan
t
a
nd
CSP
(Cloud
ser
vice
pro
vid
er
).
The
kind
of
e
nc
ryptio
n
al
gori
thm
to
be
us
e
d
is
rand
om
l
y
sel
ect
ed.
First
t
he
rec
ord
is
e
nc
rypted
by
te
nan
t
us
i
ng
public
key
an
d
then
enc
rypte
d
by
the
CSP
usi
ng
their
own
publ
ic
key.
The
pri
vate
key
of
the
te
nan
t
is
us
e
d
at
the
tim
e
of
decr
y
ption.
Th
e
key
pair
to
be
us
ed
for
eac
h
of
the
te
na
nt
is
diff
e
ren
t
a
nd
the
sam
e
is
stored
in
the
data
segm
ent
relat
ed
to
the
te
nan
t
con
ce
r
ned.
Th
e
DEP
T
al
gori
thm
wh
il
e
pro
vid
es
certai
n
le
vel
se
cur
it
y, the
data p
r
ocesse
d
wit
hin
the s
er
ver
can b
e sti
ll
b
e
at
ta
cked
b
y t
he
co
-
reside
nt u
s
ers
due
to
la
ck
of
tra
ffi
c
and
band
width
is
olati
on
.
An
at
ta
ck
m
odel
has
bee
n
presented
w
hich
is
based
on
a
threat
m
od
el
that ta
ke
s adva
ntage o
f
Mult
i
-
Te
na
nc
y si
tu
at
ion
is
presente
d by [
3].
Mi
ti
gating
the
at
ta
cking
is
the
best
cou
r
se
of
act
ion
.
T
he
inf
or
m
at
ion
relat
ed
to
resour
c
e
al
locat
ion
,
resou
rce
util
izati
on
a
nd
acce
ssing
can
be
know
n
from
the
logs
m
ai
ntained
by
the
cl
ou
ds
.
The
sca
nn
i
ng
of
t
he
log
s
an
d
a
pply
ing
brute
f
or
ce
m
e
tho
ds
th
e
det
ai
ls
of
lo
cat
ion
s
w
her
e
data
is
sto
re
d
cou
l
d
be
known
a
nd
therefo
re
ca
n
be
at
ta
cke
d.
[
4]
Hav
e
s
how
n
the
kind
of
iss
ues
t
hat
m
us
t
be
a
ddresse
d
wh
e
n
m
uti
-
te
na
ncy
is
i
m
ple
m
ented
i
n
IaaS
la
ye
r.
Wh
e
n
ne
w
ha
r
dw
a
re
is
add
e
d
with
an
inte
ntion
of
inc
rea
sing
the
pe
r
for
m
ance,
so
m
et
i
m
es it
le
ads
t
o
m
any o
f
the sec
ur
it
y i
s
su
es as
w
el
l.
The
a
uthor
s
ha
ve
pr
ese
nted
a
m
od
el
us
in
g
w
hich
t
he
pe
rfor
m
ance
of
a
cl
oud
ca
n
be
com
pu
te
d.
Secu
rity
of
th
e
data
sto
red
on
t
he
cl
ou
d
c
om
pu
ti
ng
syst
e
m
can
be
ac
hi
eved
t
hroug
h
i
m
ple
m
enting
acce
ss
con
t
ro
l
syst
em
s
co
ns
ide
rin
g
bo
t
h
a
uth
e
ntica
ti
on
a
nd
a
uthorizat
i
ons;
the
y
hav
e
prese
nt
ed
a
m
echan
i
sm
to
encr
y
pt
the
data
base
d
on
t
he
locat
ion
of
th
e
us
er
an
d
geol
ocati
on
of
the
data
w
he
re
it
has
been
store
d.
[5
]
Hav
e
prese
nted
a
c
om
par
iso
n
of
the
at
trib
ute
base
d
e
ncry
ption
(AES)
of
t
he
data
t
o
be
sto
re
d
in
th
e
cl
oud.
Howe
ver
the
m
et
ho
ds
will
be
directe
d
t
oward
s
ac
hieving
the
acce
ss
co
nt
ro
ll
in
g
of
the
da
ta
than
deali
ng
with
issues
relat
ed
t
o
m
ulti
-
te
nan
c
y.
Data
pri
vacy
a
nd
secu
rity
of
the
data
st
or
e
d
in
cl
oud
can
be
achieve
d
t
hroug
h
im
ple
m
e
nting
acce
ss
con
t
ro
l
m
echan
i
sm
s.
A
com
par
iso
n
of
c
urre
ntly
existi
ng
A
ES
-
Ba
se
d
sc
he
m
es
of
data
ac
cess
co
ntr
ol
ha
s
be
e
n
pr
ese
nted
[
6].
A
li
st
of
unsol
ved
pro
blem
s
thr
ough
AES
has
bee
n
e
nlist
ed.
E
ve
n
t
hough
t
he
AES
bas
e
d
current
e
xisti
ng
co
ntr
ol
sc
he
m
es
cou
ld
sat
isfy
the
re
quire
m
ents
of
data
a
ccess
co
ntr
ol
f
or
cl
oud
st
or
a
ge
,
there
are
sti
ll
pr
oble
m
s
su
ch
as
re
vo
cat
io
n
of
th
e
us
er
,
re
du
ct
i
on
of
th
e
com
pu
ta
ti
onal
effo
rt,
i
m
ple
m
entat
ion
of
hierar
c
hical
str
uctu
re
of
the user
et
c.
T
her
e
a
re
m
any
pr
oble
m
s
su
ch
as
re
vocat
ion
of
the u
sers
,
c
om
pu
ta
ti
on
al
eff
ic
ie
ncy,
hierar
c
hical
struc
ture
of
the
use
rs
et
c.,
w
hic
h
are
relat
ed
t
o
secu
rin
g
the
cl
ou
d
st
or
a
ge
wh
il
e
at
tribu
te
base
d encry
ption co
ul
d
s
olv
e m
any access co
ntr
ol
relat
ed
iss
ues wit
h resp
ect
t
o cl
oud
st
or
a
ge.
An
a
naly
sis
of
data
stora
ge
i
n
cl
oud
c
om
pu
ti
ng
a
nd
the
kind
of
sec
ur
it
y
enf
orcem
ent
that
can
be
bu
il
t
into
cl
oud
com
pu
ti
ng
s
yst
e
m
has
bee
n
pro
vid
e
d
[
7].
They
hav
e
em
ph
a
siz
ed
that
the
m
a
in
con
ce
pt
is
to
pro
vid
e
i
ntegr
i
ty
to
the
cl
ou
d
sto
ra
ge
area
with
disti
nct
data
m
od
el
s
a
nd
secu
rity
al
go
rithm
s.
The
y
hav
e
pr
ese
nted
cl
ou
d
data
sto
rag
e
arch
it
ect
ure
al
ong
with
t
he
cl
oud
data
m
od
e
ls.
Ma
njin
der
Singh
et
al
.
,
[
7]
hav
e
e
m
ph
asi
zed
th
at
on
e
has
to
e
ns
ure
data
inte
gr
it
y
to
the
cl
oud
st
or
a
ge
thr
ough
use
of
d
if
f
eren
t
data
m
od
el
s
and
secur
it
y
al
gorithm
s.
They
have
pr
ese
nted
a
n
arch
it
ect
ure
that
inclu
des
th
e
secur
it
y
m
od
el
s
within
the
cl
ou
d
com
pu
ti
ng
sys
tem
.
They
ha
ve
pr
es
ente
d
a
m
od
ifie
d
RSA
base
d
al
gorit
hm
that
has
be
en
prov
i
ded
with
a
diff
e
re
nt k
ey
ge
ner
at
io
n
a
nd
decr
y
ption sy
stem
f
or en
surin
g
the
clo
ud sto
rag
e
secu
rity
.
The
c
halle
ng
e
s
that
on
e
has
to
face
in
prov
i
ding
the
se
cur
it
y
within
t
he
cl
oud
c
ompu
ti
ng
a
r
e
pr
ese
nted
i
n
de
ta
il
ed
by
Kati
e
Wood
et
al
.,
[8
]
.
They
ha
ve
fo
c
us
se
d
sp
e
ci
fical
ly
aro
und
cl
oud
de
pl
oym
ent
and
data
sto
ra
ge,
in
par
ti
cula
r
rela
ti
on
t
o
pr
ivacy
con
ce
r
ns
du
e
to
m
ulti
-
te
nan
cy
.
Kati
e
Woo
d
et
al
,
[8]
hav
e
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Securi
ng
SAA
S servi
ce u
nder
cl
oud
c
omp
uting b
as
e
d m
ulti
-
te
nancy
syste
ms
(
JK
R
Sastry
)
69
pr
ese
nted
se
ri
es
of
c
halle
nges
that
on
e
m
us
t
face
w
he
n
secur
it
y
to
the
cl
oud
stora
ge
has
to
be
e
ns
ure
d.
They
hav
e
c
oncent
rated
m
ai
n
on
sec
ur
i
ng
the
cl
oud
sto
rag
e
co
ns
i
der
i
ng
the
m
ul
ti
-
t
enan
cy
im
ple
m
ented
thr
ough Saa
S.
3.
INV
E
STIG
A
TION
A
ND F
INDIN
GS
Ma
ny
appro
ac
hes
ha
ve
bee
n
pr
ese
nted
in
t
he
li
te
ratur
e
for
secur
in
g
the
da
ta
in
a
sing
le
database
in
wh
ic
h
the b
el
ongi
ng
to dif
fere
nt cu
stom
ers
has
bee
n
sto
re
d
an
d
proce
sse
d
by a sing
le
a
pp
li
cat
io
n.
A
ll
o
f
them
su
f
fer
from
one
kind
of
risk
are
ot
her.
The
data
store
d
i
n
the
data
base
c
an
be
double
e
ncr
y
pted
by
th
e
us
er
and the
n
the
se
rv
ic
e
pro
vid
e
r so as
to g
uar
a
nt
ee the c
onfide
ntial
it
y fr
om
b
oth
pe
rsp
ect
i
ve
s.
The
n
in
th
at
case
the
way
t
he
encr
y
ption
al
gorithm
s
are
sel
ect
ed
or
the w
a
y
the
keys
are
g
ene
rated
is
the
m
os
t
cru
ci
al
aspect
of
sec
ur
i
ng
the
data
store
d
in
the
da
ta
base.
Data
i
so
la
ti
on
is
m
os
t
sign
ific
a
nt
as
pect
of
secur
i
ng
t
he
da
ta
.
The
data
as
long
as
it
is
in
sing
le
data
base,
will
be
i
ns
ec
ur
e
at
le
ast
at
ph
ysi
cal
st
or
a
ge
le
vel.
Ph
ysi
cal
data
isolat
ion
can
be
done
by
distrib
uting
the
data
into
diff
e
re
nt
stora
ge
areas
co
nne
ct
ed
to
diff
e
re
nt
physi
cal
m
achines.
The
database
is
assig
ned
to
physi
cal
data
st
or
a
ge
sit
uate
d
in
dif
fe
ren
t
m
achines
.
Database
ca
n
be
portio
ne
d
horiz
on
ta
ll
y
and
each
par
ti
ti
on
is
m
ade
to
be
sit
uated
on
th
e
storag
e
sit
ua
te
d
in
diff
e
re
nt
m
achines.
Eac
h
pa
rtit
ion
is
al
lott
e
d
on
e
sin
gle
use
r.
Since
t
he
data
of
dif
fer
e
nt
use
rs
is
sit
ua
te
d
in
diff
e
re
nt
m
achines,
data
isol
at
io
n
is
achiev
ed
there
by
there
is
no
cha
nc
e
of
one
us
er
encroac
hing
in
to
the
stora
ge
a
rea
of
so
m
e o
ther use
r.
T
he
a
rr
a
nge
m
ent o
f
su
c
h
a
n
im
ple
m
entat
i
on is s
how
n
in
Figure
1.
Figure
1
.
Mult
i
-
portio
ning t
he
d
at
aba
se
Each
of
the
pa
rtit
ion
is
recog
nized
by
the
I
P
address
of
the
Ma
chine
w
her
e
it
is
sit
ua
te
d
and
th
e
locat
ion
where
the
pa
rtit
ion
is
sit
uated.
A
use
r
id
is
ide
ntifie
d
by
an
I
D
assigne
d
by
th
e
Cl
oud
com
pu
ti
ng
so
ft
war
e
.
Wh
e
n
a
us
er
m
akes
a
req
uest
f
or
a
VM
to
ru
n
a
S
aaS
serv
ic
e
the
hyperviso
r
sha
ll
m
ake
a
request
to
the
ap
plica
ti
on
to
al
locat
e
the
par
ti
ti
on
an
d
pro
vid
e
the
det
ai
ls
of
the
Ph
y
sic
al
m
achine
(I
P
a
ddres
s)
a
nd
the
locat
ion
w
her
e
the
par
ti
ti
on e
xists withi
n
the
physi
cal
m
ach
ine.
The
use
r
ca
n
encr
y
pt
his
own
data
us
i
ng
their
public
key.
Th
e
publ
ic
key
of
the
us
er
ca
n
be
gen
e
rated
on
t
he
us
er
side.
The
al
go
rithm
to
be
us
e
d
by
the
us
e
r
f
or
unde
rtakin
g
t
he
e
ncr
ypti
on
can
be
fetche
d
dynam
ic
al
ly
at
the
us
er
side
ba
sed
on
his
I
D.
T
he
key
an
d
t
he
enc
ryptio
n
a
lgorit
hm
are
use
d
to
encr
y
pt
the
data
o
n
the
us
e
r
s
ide
befor
e
it
is
transm
i
tt
ed
to
the
hyper
wis
er
for
stori
ng
t
he
sam
e
on
th
e
us
er
relat
ed
portio
n.
The
hype
rv
is
or
sh
al
l
lookup
f
or
an
al
gorith
m
based
on
the
IP
address o
f
the
physi
cal
m
a
chine
and
t
he
locat
io
n
of
the
pa
rtit
ion
within
the
physi
cal
m
achine,
us
in
g
w
hic
h
an
enc
ryptio
n
al
gorithm
and
a
key
is
ge
ner
at
e
d
a
nd
the
sam
e
are
us
e
d
f
or
e
ncr
y
pting
data
on
t
he
cl
ou
d
side.
T
hu
s
th
e
data
protect
ion
is
i
m
ple
m
ented
and
pr
i
vacy
m
a
intai
ned
on
bo
th
the
en
ds
of
the
cl
ie
nt
and
CSP.
T
he
des
cripti
on
of
the
data
howe
ver can
be u
nd
e
rtake
n o
n
the
cli
ent sid
e u
si
ng the
pr
i
vate k
ey
of the
u
se
r.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
1
,
Ja
nu
a
ry
201
9
:
6
5
–
7
1
70
Algorithm
In
it
ia
li
zat
ion
pro
ces
s
a)
Cl
ie
nt r
eq
uest
for parti
ti
on in t
he
data
base.
T
he req
uest se
nt
to
the
H
y
pervi
so
r
.
b)
The
hype
rv
is
or
gets
t
he
deta
il
s
of
the
pa
rtit
ion
in
te
rm
s
of
t
he
I
P
ad
dr
ess
an
d
the
l
oc
at
ion
of
t
he
par
ti
ti
on
fr
om th
e guidin
g
ap
pl
ic
at
ion
so
ftwa
re.
A
dif
fer
e
nt I
P addr
es
s is c
ho
s
en
eve
ry ti
m
e a
r
equ
es
t
is i
niti
at
ed.
c)
The
ke
y
and
t
he
enc
ryptio
n
al
gorithm
t
o
be
us
ed
on
the
cl
oud
side
a
r
e
gen
e
rated
ba
sed
on
the
TCP/
IP
a
ddres
s and t
he
locat
i
on of t
he parti
ti
on
.
d)
The use
r
I
D, K
ey
an
d t
he
e
nc
r
ypti
on
al
gorith
m
are
stored along
with
us
e
r I
D wit
hin
t
he
cl
oud.
Cl
ie
nt side pr
oc
ess
a)
Tena
nt
‘Ci’
ge
ner
at
es
a
la
rg
e
Pr
im
e
Cp
fr
om
his
ow
n
cre
den
ti
al
s
ge
ner
a
ll
y
us
ing
his
own
I
D
a
nd
sent to
Cl
oud S
erv
ic
e
Prov
i
der.
b)
Tena
nt Ci com
pu
te
s
N
=
2*Cp
c)
Tena
nt Ci ge
ne
rates Cy
cl
ic
g
r
oup ZN
* of o
r
der Ø
(N)(
Euler
Quo
ti
ent
f
un
ct
ion)
.
d)
A
s
ubgro
up Z
Ø(N)* s
ubset
of Z
N* of
orde
r Ø(
Ø(N)) is
ge
ne
ra
te
d by Ci
w
i
th g
e
ne
rator g
∈
Zn
*
e)
Tena
nt
Ci
ra
ndom
l
y
picks
up
two
pri
vate
ke
ys
Tq
a
nd
C
r
∈
ZN
*
C
q≡
gk
1
m
od
N
a
nd
Cr≡
gk
2
m
od
N wh
e
re
k1,
k2
∈
Z
Ø(N)*
w
he
re
g
is
gen
e
rat
or for
ZN*
f)
Tena
nt Ci com
pu
te
s
N
=
Cq
* C
r
g)
Ci
ch
ooses
‘
e
’ suc
h
that
gc
d(
e
, Ø(N))
=
1.
h)
Ci
d
et
erm
ines ‘
d’ s
uc
h
that e
d≡ 1 m
od
Ø(N
)
i)
Tena
nt
Ci
com
pu
te
s
CP
r
=
e.
rst
su
c
h
that
e.
rst
≡
1
m
od
Ø
(N)
an
d
CP
b
=d.rsd
s
uc
h
th
at
d.
rs
d
≡
1
m
od
wh
e
re
C
Pr
:
T
ena
nt
P
r
ivate
Key,
CP
b:
Te
nan
t
publ
ic
key
P
ubli
c
key
<N
,CP
b>
Pr
ivate
key
<C
Pr
,
d,
e>
j)
Tena
nt Ci enc
r
ypts t
he
data
of each
rec
ord
R
(ER) ER
= Re
m
od
n
k)
Tena
nt Ci sen
ds ER
j t
o
C
S
P t
o
st
or
e i
n
it
s P
arti
ti
on
Pi
.
Pr
oc
essin
g o
n
t
he
Cl
ou
d
side
a)
CSP f
et
c
hes
t
he
r
el
at
ed Enc
ryption al
gorith
m
an
d
the
key
with the
h
el
p o
f user
ID.
b)
The dat
a r
eco
r
d recei
ve
d
f
r
om
the cli
ent is
encr
y
pted
a
gai
n usin
g
t
he key
and e
ncr
y
ptio
n
al
go
rithm
c)
EER = ER
TP
b m
od
n
d)
Fetc
h
the
p
a
rtit
ion
detai
ls PT
of the cli
ent
usi
ng
t
he
lo
okup
ta
ble w
it
h
t
he help
of
us
er
ID
e)
CSP sto
res E
R
in p
a
rtit
ion
Pi
of Ci
Data Re
trie
val
process
a)
Tena
nt
encr
y
pts
the
Pr
im
ary
data
us
in
g
his
own
key
a
nd
a
lgorit
hm
and
sends
the
sam
e
to
the
cl
oud
al
ong wit
h his
own ID
.
b)
CSP
fetc
hes
t
he
rec
ord
from
t
he
par
ti
ti
on
rel
at
ed
to
t
he
cl
ie
nt
usi
ng
the
en
crypted
key
da
ta
wh
ic
h
is
furthe
r
encr
y
pted
us
in
g
the
cl
ie
n
t’s
encr
y
ption
al
gorithm
a
nd
the
key
w
hi
ch
is
gen
erate
d
and
store
d
in
the
look
up
t
able
on
the
CS
P
side.
T
he
det
ai
ls
of
the
pa
rtit
ion
relat
ed
to
t
he
cl
ie
nt
are
store
d
withi
n
the Ap
plica
ti
on
s
of
t
war
e
or t
he hype
rv
is
or.
c)
The
que
ried
da
ta
is
sent
to
the
cl
ie
nt
.
It
sh
oul
d
be
no
te
d
that
no
decr
y
ption
is
done
on
the
cl
oud
side
.
Qu
e
ryi
ng is
done usin
g
t
he
e
nc
rypted
k
ey
va
lues
on
ly
.
d)
Af
te
r
r
ecei
ving
Ten
a
nt Ci c
om
pu
te
s R = EE
Rrst m
od
N to
ob
ta
in
origi
nal
Re
cord.
e)
If
Te
nan
t
Ci
do
es
not
get
Re
cord
R
from
a
bove
data
the
n
Ci
assu
m
es
R
is
m
od
ifie
d
by
CSP
or
intruder
, so R i
s d
isc
a
rd
e
d
a
nd
requests
for f
resh rec
ord.
4.
RESU
LT
S
AND A
N
ALYSIS
The
a
bove
m
e
ntion
al
gorit
hm
has
bee
n
im
plem
ented
within
the
Eucaly
ptu
s
an
d
eve
n
t
he
brute
force
m
et
ho
d
a
ppli
ed
to
acce
ss
t
he
data
sit
ing
o
n
the
data
base
s
of
t
war
e
si
de
did
no
t
rev
eal
t
he
secrecy
of
t
he
data
store
d
in
the
da
ta
base.
It
has
no
t
bee
n
poss
ible
to
even
lo
cat
e
the
par
ti
tio
n
or
ga
in
ha
ndle
on
the
al
gorithm
and the
key
use
d for
unde
rtak
ing
t
he
e
ncr
y
ption
on the
CSP
side.
5.
CONCL
US
I
O
N
Wh
e
n
ap
plica
ti
on
s
of
twa
re
t
ha
t
us
es
a
datab
ase
has
to
be
pro
vid
e
d
a
s
data
ser
vice
to
m
ulti
ple
us
e
rs
thr
ough
virtu
al
m
achines,
the
issue
of
Mult
i
-
te
nan
cy
arises.
The
data
can
be
at
ta
cked
by
the
us
ers
due
to
the
reasons
of
m
ulti
-
te
nan
cy
.
One
us
er
ca
n
at
ta
ck
ot
her
as
both
the
VM
relat
ed
to
the
cl
ie
nts
are
sit
uated
on
th
e
sam
e
serv
er.
T
her
e
fore
it
beco
m
es
necessary
to
i
m
ple
m
ent
m
et
ho
ds
/
Me
chan
ism
s
that
help
in
pr
otect
ing
th
e
data
w
hen
t
he
sam
e
serv
ic
es
are
pro
vid
e
d
to
sever
al
us
ers
.
Data
isolat
ion
is
the
key
t
o
pr
otect
the
data
wh
ic
h
can
be
ac
hiev
e
d
t
hroug
h
l
oca
ti
on
s
a
nd
t
he
pa
rtit
ion
s
of
data
relat
ed
to
a
s
pecific
us
er
on
a
diff
e
re
nt
ph
ysi
cal
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Securi
ng
SAA
S servi
ce u
nder
cl
oud
c
omp
uting b
as
e
d m
ulti
-
te
nancy
syste
ms
(
JK
R
Sastry
)
71
m
achine
or
in
a
com
plete
ly
diff
ere
nt
locat
i
on
on
the
sam
e
physi
cal
m
ac
hin
e.
Mo
re
c
om
plex
bu
t
sim
ple
to
i
m
ple
m
ent
sys
tem
can
be
a
chieve
d
th
r
ough
do
ub
le
e
nc
ryptio
n
car
rie
d
on
t
he
c
us
t
om
er
and
cl
oud
side.
The key
a
nd th
e enc
ryptio
n
al
gorithm
are
sel
ect
ed
dy
nam
icall
y using the
r
efere
nce
data.
REFERE
NCE
S
[1]
M.Sara
sw
at
hi
,
Dr.T
.
Bhuva
nes
wari
,
“
Multi
te
n
ancy
in
Cloud
Software
as
a
S
erv
ic
e
Appli
cation”,
Inte
rnatio
nal
Journal
of
Ad
va
nce
d
Re
search
i
n
Computer
Sc
ience
and
Sof
tware
Eng
ine
ering
,
v
ol3,
issue11,
pag
e
1
-
4,
2013
.
[2]
K.Venka
taram
an
a,
Prof.
M.
Pad
m
ava
tha
m
m
a,
“
Multi
-
Te
n
ant
D
a
ta
Storag
e
Secu
r
ity
In
Cloud
Us
ing
Data
Par
ti
t
io
n
Enc
r
y
pt
io
n
T
ec
h
nique
”
.
In
te
rnati
onal
Journal
of
Sci
en
ti
fic
&
Eng
ine
ering
Re
se
arc
h
,
vol4
,
issue7,
p
a
ge1
-
5,
2013.
[3]
Hus
sain
AlJahda
li
,
Abdula
ziz
Al
bat
li,
Pet
er
Garr
agha
n,
Pau
l
Tow
nend,
L
y
d
ia
L
au
,
Jie
Xu,
Multi
-
T
ena
nc
y
in
Clou
d
Com
puti
ng,
IEE
E
8th
Inte
r
nati
onal
Symp
osium
on
Ser
vi
c
e
Or
ie
nte
d
Syste
m
Engi
nee
ring
(
SOSE
)
,
doi.
org/10
.
1109.
SO
SE.
2014.
50,
p
age
1
-
9,
2014
.
[4]
Bhawna
Sehga
l
Er.
Jasb
ee
r
N
ar
wal,
“
An
Anal
ysis
of
Perform
a
nce
for
Mult
i
-
T
ena
nt
Applicat
io
n
through
C
lou
d
SIM
”,
Inte
rnat
io
nal
Journal
of
E
merging
Re
searc
h
in
Manag
eme
n
t
&
T
ec
hnology
,
v
ol4
issue6,
p
age
1
-
5,
2015.
[5]
Goikar
Vanda
n
a
T.
,
Jagda
l
e
Sup
ri
y
a
K.
,
Para
d
e
Pri
y
a
B.
,
Pawar
Sum
edha
D.,
“
Im
prove
Secur
ity
of
Data
Ac
ce
ss
in
Cloud
Com
puti
n
g
using L
o
cation
”,
I
JCSMC
,
vol4
issue2,
page
1
-
1
0,
2015.
[6]
Te
ngfe
i
Li,
Lian
g
Hu,
Yan
Li
,
Ji
anf
en
g
Chu,
Ho
ngtu
Li
,
and
Hong
y
ing
Han
,
“
Th
e
Resea
r
ch
and
Pros
pec
t
of
Secu
re
Data
Acc
ess
Co
ntrol
in
Cloud
Storage
Env
iron
m
ent
”,
Journal
of
Comm
unic
ations
,
VO
L
10,
IS
SU
E
10,
PA
GE
1
-
7,
2015.
[7]
Manji
nder
Sing
h,
Chara
n
ji
t
Si
ngh,
“
Multi
T
e
nancy
Secur
ity
in
Cloud
Com
puti
ng”
,
Inte
rnat
i
onal
Journal
O
f
Engi
ne
ering
Sc
i
enc
es
&
Re
searc
h
Technol
og
y
,
V
ol
4,
Iss
ue
116,
P
age
1
-
7
,
2017.
[8]
Kati
e
W
ood
a
nd
Dr
Mark
Anderson,
“
Understa
nding
the
complexi
t
y
su
rrounding
Multi
te
nan
c
y
in
cl
o
ud
computing”
,
E
ig
hth
IE
EE Int
ern
ati
onal
Con
fe
ren
ce
on
e
-
B
usiness E
ngin
ee
ring
,
10
.
1109/ICE
B
E.
20
11.
68,
2011
.
[9]
htt
p://ww
w.ga
rt
ner
.
com/id
=205
8722.
[10]
K.
W
ood,
M.
A
nder
son,
“
Under
standi
ng
th
e
co
m
ple
xity
surrou
nding
m
ult
itena
nc
y
in
cl
oud
co
m
puti
ng”,
Ei
ght
h
IEE
E
Inte
rnat
io
nal
Conf
ere
nce
on
e
-
Busine
ss
E
ngine
ering
,
VO
L1,
PA
GE
NO
119
-
124,
2011.
[11]
Paul
Fere
st
en, “S
tora
ge
Mul
ti
-
T
ena
nc
y
for
C
lou
d
Com
puti
ng”,
S
NIA
,
2010.
[12]
W
.
Tsai
,
Q.
Shao,
“
Role
-
Bas
ed
Acc
ess
-
Cont
rol
Us
ing
Refe
ren
ce
Ontolog
y
in
Clouds”,
Tenth
Inte
rnat
ion
al
Symposium on
A
utonomous Dec
e
ntrali
zed
S
yste
m
s
,
VO
L
11,
PA
GE
121
-
128
,
2011
.
[13]
Moham
ed
Almors
y
,
John
Grund
y
,
and
Am
ani
S.
Ibra
him,
“
TOS
SM
A:
A
Te
nant
-
Ori
ent
e
d
SaaS
Secur
i
t
y
Mana
gement
Ar
chi
t
ec
tur
e”,
I
EEE
F
if
th
Int
ernational
Conf
ere
nce on
Cloud
Computing
,
PA
GE
1
-
9
,
2012.
[14]
S.Kam
ara
,
Kristi
n
La
ute
r
,
“
Cr
y
pt
ogra
phic
c
loud
storage”,
FC
'10
Proce
ed
ings
of
the
14th
int
ernat
i
onal
conf
ere
n
c
e
on
Fi
nan
ci
al
cry
ptography
and
d
ata
sec
uri
ty
,
PA
GE
136
-
149,
201
0.
[15]
Jos
e
M.
Alca
ra
z
Cal
ero
,
Nige
l
Edwa
rds,
Johan
nes
Kirschnic
k,
La
wrenc
e
W
il
c
ock,
and
Mik
e
W
ra
y
,
“
Towa
rd
a
m
ult
i
-
te
n
ancy
au
thori
z
at
ion
s
y
st
e
m
for
cl
oud
serv
ic
es”
,
I
EE
E
Se
cu
rity
and
Priv
a
cy
,
PA
GE
48
-
55,
20
10.
Evaluation Warning : The document was created with Spire.PDF for Python.