Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
6, N
o
. 2
,
A
p
r
il
201
6, p
p
.
70
8
~
71
6
I
S
SN
: 208
8-8
7
0
8
,
D
O
I
:
10.115
91
/ij
ece.v6
i
2.8
270
7
08
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
A M
o
del
Dri
v
en Fram
ework for Portabl
e
Cloud S
e
rvi
ces
A
p
a
r
na
V
i
j
aya
*, N
eela
n
a
r
ay
anan V
*
* School of
Com
puter Science
an
d Engin
eering
,
Vello
re Institu
te
of Technolog
y
C
h
ennai
Cam
pus,
India
Article Info
A
B
STRAC
T
Article histo
r
y:
Received J
u
n
4, 2015
Rev
i
sed
No
v
28
, 20
15
Accepted Dec 17, 2015
Cloud Computing is an evolving
technolog
y
as it offers significant benefits
like p
a
y onl
y for
what
you us
e
,
s
cal
e th
e res
our
ce
s
accord
ing to
th
e ne
eds
and
less in-house staff and r
e
sources. These
b
e
nef
its
have r
e
sulted
in
tremendous
increase in the n
u
mber of applications
and services hosted in the cloud which
inturn has resulted in incr
ease in the number of cloud providers in the
market. C
l
oud s
e
rvice provid
e
rs have
a lo
t of h
e
terogen
e
ity
in
th
e resources
they
use. Th
ey
h
a
ve th
eir
own server
s, diff
eren
t cloud infrastru
ctu
r
es, API’s
and methods to access the clo
ud resour
ces. Despite its benefits; lack o
f
standards among service provid
e
r
s
has
caused a high level of vendor lock-in
when a s
o
ftware
develop
e
r tri
e
s
t
o
change
its clou
d provider. In
this paper we
give
an ov
erview on the ongoing and
curren
t
trends in
the ar
ea of
clou
d
service portability
and we also pr
opos
e a new cloud portability
p
l
atform. Our
new platform is based on establishi
ng f
eatur
e models which offers the
desired cloud p
o
rtability
.
Our soluti
on DSky
L uses feature models and
domain model analy
s
is
to support
develo
pment,
customization and
deplo
y
m
e
nt of application com
p
onents across m
u
ltipl
e
clouds.
T
h
e m
a
in goal
of our approach is to reduce the effor
t
and
time need
ed
for porting
applications acr
oss different clo
uds. Th
is paper aims to give an overview on
DSky
L.
Keyword:
Feat
ure
M
o
del
Het
e
r
oge
n
ous
cl
ou
d
M
odel
dri
v
en
devel
opm
ent
Platform
agn
o
s
tic
Copyright ©
201
6 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Apa
r
na Vijay
a
Sch
ool
o
f
C
o
m
put
e
r
Sci
e
nce a
n
d
E
ngi
neeri
n
g
,
V
e
llo
re In
stitu
te of Tech
no
logy Ch
enn
a
i Cam
p
u
s
,
Ch
enn
a
i 6
001
27
, Tam
i
l
N
a
du
,
In
d
i
a.
Em
a
il: ap
arn
a
v8
5@g
m
ai
l.co
m
1.
INTRODUCTION
Soft
ware i
n
d
u
s
t
ry
i
s
now i
n
a
t
r
ansi
t
i
on p
h
a
s
e from
deskt
o
p com
put
i
ng t
o
cl
ou
d com
p
u
t
i
ng. C
l
o
u
d
com
put
i
ng of
f
e
rs
se
veral
be
nefi
t
s
i
n
cl
udi
n
g
vi
rt
ual
i
zed
har
d
ware
i
n
f
r
a
st
ruct
u
r
e, use
r
sel
f
pr
o
v
i
s
i
oni
ng
,
el
ast
i
c
i
t
y
and pay
per u
s
e. I
n
t
h
i
s
era of cl
ou
di
fi
cat
i
o
n t
h
ere i
s
enorm
o
u
s
num
ber of cl
ou
d p
r
o
v
i
d
e
r
s
i
n
t
h
e
mark
et. Th
e
mark
et k
e
ep
s
m
o
v
i
n
g
o
n
and
th
e ab
ility to
su
rv
iv
e in
a m
a
rk
et crow
ded
w
ith
cloud
v
e
ndors
becom
e
s a tedious tas
k
. E
v
e
r
y cloud
ve
ndor try to introduc
e
flexibility in their offe
ring
s
so as to a
d
a
p
t to the
mark
et and to surv
i
v
e. Th
is
flex
ib
ility creates a lo
t
of con
f
u
s
ion am
o
n
g
th
e
u
s
ers and th
ereb
y cho
o
s
in
g a
cl
ou
d p
r
o
v
i
d
e
r
becom
e
s
m
o
re or l
e
ss a g
u
es
si
ng
gam
e
. User nee
d
s m
i
ght
be sat
i
s
fi
ed at
i
n
cept
i
o
n by
a
cl
ou
d
v
e
ndo
r,
b
u
t
u
s
er
n
eed
s ch
ange b
a
sed
on
their
bu
sin
e
ss escalatio
n
s
.
A
t
this po
in
t sw
itchin
g
cl
ou
d pr
ovid
e
r
s
can c
o
m
e
at a cost. T
h
e
r
e m
i
ght
be
hidde
n
l
o
ck-in cost
s
wi
th a s
p
ecific cl
oud
provid
e
r
or proprietary s
e
rvice
m
odel
such
as
t
h
e c
o
st
of
m
ovi
ng c
ode
and
dat
a
fr
om
o
n
e cl
o
u
d
se
rvi
ce
pr
o
v
i
d
e
r
t
o
a
not
her
m
i
ght
be
p
r
oh
ib
itiv
e. It
is n
o
t
on
ly m
o
v
i
ng
in
to
and
o
u
t
o
f
a cloud
en
v
i
ron
m
en
t a co
stly en
d
e
avo
r
, bu
t task
s su
ch
as
refactoring
th
e co
d
e
can
result in
u
n
e
xpecte
d
costs.
In s
o
me scenarios,
t
h
i
s
swi
t
c
h nee
d
ed m
a
jor c
h
a
nge
s i
n
soft
ware
an
d c
a
use
d
pr
oject
d
e
l
a
y
s
and
eve
n
pr
o
duct
i
v
i
t
y
l
o
sses.
C
l
ou
d com
put
i
ng
[
1
]
,
[
2
]
ena
b
l
e
s an
o
r
ga
ni
zat
i
on t
o
de
pl
o
y
and
ru
n t
h
ei
r
appl
i
cat
i
o
ns i
n
a
pl
at
fo
rm
main
tain
ed
b
y
th
ird
p
a
rty cl
o
u
d
prov
id
ers rath
er t
h
an main
tain
in
g
t
h
eir
own
d
a
ta cen
ters. Org
a
nizatio
n
s
(cust
o
m
e
rs)
re
nt
t
h
e
res
o
urc
e
s
of
t
h
e
cl
o
u
d
, IaaS on w
h
i
c
h
ap
pl
i
cat
i
o
n
s
can be
de
pl
oy
ed or PaaS whe
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
70
8 – 7
1
6
70
9
devel
ope
rs ca
n create their
own Saa
S
applications. Final
l
y, end
user
s
access and us
e the de
ploye
d
Saa
S
ap
p
lication
s
. In
th
is scen
ari
o
o
f
cloud adoption,
users
of PaaS se
rvi
ces are raisin
g
th
e q
u
e
stio
n o
f
th
e
p
o
rtab
ility o
f
th
eir app
licatio
n
s
fro
m
o
n
e
p
r
o
v
i
d
e
r to
a d
i
fferen
t
o
n
e
, o
r
ev
en
b
ack to
th
e d
a
ta cen
t
er.
Porta
b
ility is
the ability to
m
ove a
p
plications am
ong
diffe
rent platform
s without ha
vi
ng to rewrite it partly or
fu
lly. It is t
h
e
p
r
erequ
i
site for bu
ild
i
n
g tru
l
y
ag
ile and
flex
i
b
le system
s th
at d
o
no
t lo
ck
in
th
eir u
s
ers.
Du
e to
t
h
e en
orm
ous
num
ber
of cl
o
u
d
p
r
o
v
i
d
e
r
s i
n
t
h
e m
a
rket
and the i
n
crea
sing
num
b
er of services
offe
red
or
created,
portability of service
s
is now bec
o
m
i
ng a
m
a
jor conce
r
n.W
i
t
h
the cloud
becom
i
ng m
o
re com
p
etitive
an
d
so
m
e
p
r
o
v
id
ers are at lo
ng
-term
stab
ilit
y risk
s, d
e
v
e
lop
e
rs
o
r
cu
st
o
m
ers h
a
v
e
to
u
n
d
e
rstan
d
th
at ad
h
e
ring
to a s
p
ecific vendor’s
Platform
as a
Service (PaaS) ca
n
be
a risk beca
use
of
the u
n
eve
n
sup
p
o
rt fo
r platform
feat
ure
s
by
di
f
f
ere
n
t
cl
o
ud
p
r
ovi
ders
.
Whe
n
t
h
ere a
r
e m
u
l
t
ip
le op
tio
n
s
in
p
r
i
v
ate as well
as pu
b
lic cloud
, it is
a v
e
ry challeng
ing
task
to iden
tify th
e b
e
st
su
ited cloud
p
r
ov
id
er for an
o
r
g
a
n
i
zatio
n
.
Prov
id
ers
h
a
ve tried
introducing
ne
w se
rvices, ent
e
red t
h
e
ope
n s
o
u
r
ce wo
rl
d
,
m
e
rge
d
wi
t
h
gi
ant
s
in t
h
e cloud m
a
rket and s
o
on t
o
with
stand
th
e co
m
p
etitio
n
.
Th
ey h
a
v
e
tried red
u
c
i
n
g
th
e
p
r
ice as th
ey ch
allen
g
e
each
o
t
h
e
r to
see h
o
w low
th
ey can
go
. Th
is b
a
ttle led
man
y
v
e
nd
ors
fail in
th
e m
a
r
k
et.
Wh
en
we
h
a
v
e
a
d
i
v
e
rsified
clo
u
d
strat
e
g
y
it
g
i
v
e
s m
u
ltip
le
op
tio
ns so
t
h
at we can
al
ways h
a
n
d
l
e
some o
f
t
h
e
u
n
fo
reseen
risks.
So
m
e
ti
mes th
e clo
ud
v
e
ndo
rs fail to o
f
fer con
tinu
a
l clo
u
d
serv
ices. As a
resu
lt th
e ap
p
lication
th
at is ho
sted in
th
e clou
d b
e
co
m
e
s
inaccessible affecting
busine
ss conti
nuity.
If the
services
are porta
b
le
the cha
n
ces
of res
o
urce s
h
a
r
ing
b
e
tween
th
e
cl
o
u
d
s
are h
i
gh
wh
ich
will
i
m
p
r
ov
e
bu
sin
e
ss co
n
tinu
ity and
reliab
ility.
Seve
ral ope
n standards
,
APIs and t
ools have
be
en proposed [5] [6]
[7] [8] [9] [10] [11] in recent
years for ach
iev
i
ng
clo
u
d
portab
ility. A co
m
p
arativ
e stu
d
y
o
f
v
a
ri
o
u
s
ap
pro
ach
es
h
a
s b
een
d
o
n
e
in
[1
], [4
]
an
d is co
n
s
o
lidated
in
t
h
e
fo
ll
o
w
i
n
g tab
l
e.
Tabl
e
1. C
o
m
p
rat
i
v
e St
udy
of
Exi
s
t
i
n
g
Ap
p
r
oache
s
Approach
Na
m
e
Languages
supported
Data Support
OS
Clouds Tested
Vendor
Independen
t
Met
h
odology
adopt
e
d
OCCI
Java,
Ruby
,
er
lang
A
m
azon
Cross
pla
t
form
OpenNebula,
mn
e
s
i
a
Yes
Cloud-
specific
standards
Sim
p
leCloud PHP
Am
azon S3 and
Nirv
an
ix
IMF
S
Cross
pla
t
form
Z
e
nd Cloud
No
API
deltaCloud
Ruby
A
m
azon, swift,
walrus
Cross
pla
t
form
RackSpace,
OpenNebula,
AWS
,
GoGrid
No
API
jCloud java
A
m
azon
Cross
pla
t
form
RackSpace,
HpHelion,
AWS
,
GoGrid
No
API
mO
S
A
I
C
Java,
Py
thon,
er
lang,
node.
j
s
Riak,
CouchDB,
Me
m
c
achDB,
Redis, My
SQL,
Am
azon S3,
HDFS
L
i
nux
Am
azon E
C
2,
OpenNabula,
Eucalyptus,
Yes
Multiagent Syste
m
s
/tool
OASIS
TOSC
A
Java, PHP
MyS
Q
L
Linux
OpenStack
Yes
Cloud-
specific
standards
Docker
Java,
Ruby
,
Per
l
,
Py
thon,
node.
j
s
Riak, Cassandra,
M
ongoDB
L
i
nux
Azure, AWS,
OpenStack
Container
/ tool
M
ODA Clouds
Java,
Py
thon,
er
lang,
node.
j
s
Riak,
CouchDB,
Me
m
c
achDB,
Redis, My
SQL,
Am
azon S3,
HDFS
Lin
u
x
Eu
calyp
tu
s
Yes
M
odel-
dr
iven
application
engineer
ing
Openshif
t
Java,
Py
thon,
per
l
,
Ruby
,
PHP, .NE
T
M
ongoDB,
MyS
Q
L,
PostgreSQL,
M
i
cr
osoft SQL
Server
L
i
nux
-
Yes
Hybrid Platf
o
r
m
a
s
a Service
Artist
Java
MyS
Q
L
Cross
pla
t
form
- Yes
-
2.
PROP
OSE
D
APP
R
O
A
CH
Fo
r ach
i
ev
ing ap
p
lication
po
rtab
ility we h
a
v
e
u
s
ed
th
e tech
n
i
qu
e
o
f
m
o
d
e
l driv
en
software
d
e
v
e
l
o
p
m
en
t.
Mo
d
e
l
d
r
i
v
en so
ft
w
a
re
d
e
v
e
lop
m
en
t ai
m
s
to
in
crease t
h
e
v
i
sib
ility o
f
know
ledg
e co
m
p
ared to
th
e trad
ition
a
l
p
r
o
cess
b
y
exp
licit rep
r
esen
t
a
tio
n
o
f
i
n
fo
rmatio
n
th
rou
g
h
m
o
d
e
ls. It
prov
id
es kno
wl
ed
g
e
at
d
i
fferen
t
lev
e
ls o
f
ab
st
ractio
n; th
at is it separate
s th
e con
c
ep
tu
al
d
e
tails fro
m
th
e i
m
p
l
emen
tatio
n
d
e
tails
.
It
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Mo
del
Dri
v
e
n
Fr
a
m
ew
ork
f
o
r P
o
rt
abl
e C
l
ou
d
Servi
ces
(
A
pa
rna
Vija
ya)
71
0
al
so p
r
om
ot
es aut
o
m
a
t
i
c
generat
i
on
o
f
co
d
e
t
o
o
v
erc
o
m
e
t
h
e abst
ract
i
o
n ga
p
bet
w
ee
n
t
h
e m
odel
s
and t
h
e
im
pl
em
ent
a
t
i
o
n pl
at
f
o
rm
.
The
m
odels also
act as a common c
o
mm
uni
cat
i
on t
e
rm
bet
w
een st
ake
hol
de
rs an
d
b
u
s
i
n
ess du
r
i
ng
r
e
qu
ir
em
en
ts
and
d
e
sign
.
Feat
ure m
odel
i
ng i
s
i
n
depe
n
d
e
nt
of t
h
e
un
d
e
rl
y
i
ng p
r
o
g
r
a
m
m
i
ng l
a
ngua
ge. H
o
weve
r
we ha
ve use
d
Jav
a
as end
user p
r
og
ramm
i
n
g
langu
ag
e. Th
e m
a
in
reaso
n
for cho
o
sin
g
Jav
a
is th
e g
r
eater flex
ib
ility
co
m
p
ared
to
a cu
sto
m
d
o
m
ai
n
sp
eci
fic lan
g
u
a
g
e
(DSL) and
it also
i
m
p
r
ov
es reu
s
ab
ility
o
f
d
o
m
ain
o
b
jects.
DS
ky
L’s
[1]
[
4
]
use
r
i
n
t
e
r
f
a
ce has t
h
ree
p
a
nel
s
- t
h
e
di
agram
edi
t
o
r, t
h
e
pr
ope
rt
y
pa
nel
an
d t
h
e
pa
ckag
e
ex
p
l
o
r
er
p
a
n
e
l. Th
e d
i
ag
ram
ed
ito
r d
i
sp
lays th
e
d
i
agram
an
d all th
e elem
en
ts it co
n
t
ain
s
, and allo
ws the u
s
er
t
o
d
r
ag
an
d
dr
op
ne
w
feat
u
r
e
s
an
d rel
a
t
i
o
ns
hi
ps i
n
t
h
e
edi
t
or
. T
h
e
pr
ope
r
t
i
e
s panel
al
l
o
ws
user
s t
o
vi
e
w
a
n
d
m
odify the prope
rties of every ele
m
ent. Packa
g
e E
xpl
ore
r
dis
p
lays the ele
m
ent location
or element
hierarc
h
y.For each ele
m
ent which is dr
oppe
d into the dia
g
ram
panel, the st
ring
give
n in the property pa
nel is
interpreted, and the
gra
phical
represen
tation o
f
th
at elem
en
t is
m
o
d
i
fied
acco
rd
ing
to
t
h
is in
terpretatio
n. Th
is
i
s
achi
v
ed
usi
n
g ja
va re
fl
ect
i
ons AP
I. T
h
e c
ode i
s
co
m
p
iled
con
tin
uou
sly in
th
e
background, and immediate
feed
bac
k
i
s
pr
o
v
i
d
e
d
t
o
t
h
e en
d use
r
by
re
dra
w
i
n
g t
h
e el
em
e
n
t. If the c
o
de contains errors
, the affecte
d
source
code
l
i
n
es a
r
e
hi
g
h
l
i
ght
e
d
.
The t
o
pol
ogy
of t
h
e a
ppl
i
cat
i
on i
s
re
prese
n
t
e
d i
n
term
s
of feat
ures as
a Feature Model. Feature
m
odel
represe
n
t
s
a t
r
ee w
h
o
s
e no
des a
r
e t
h
e ap
pl
i
cat
i
on
com
pone
nt
s o
r
feat
ures
, an
d
wh
ose e
dge
s are t
h
e
relatio
n
s
b
e
tween
th
ese app
licatio
n
co
mp
on
en
ts as specified
in
Fi
gu
re
1
.
In
software
p
r
o
d
u
c
t
lin
e
engi
neeri
n
g
,
o
n
e ca
n co
n
f
i
g
u
r
e t
h
e
pr
o
duct
by
sel
ect
i
ng t
h
e desi
re
d
feat
u
r
es f
r
o
m
a feat
ure m
odel
bas
e
d
on
cu
sto
m
er’s fu
nctio
n
a
l requ
iremen
ts. Th
e commo
n
a
lities a
n
d
v
a
riation
s
are id
en
tified
du
ri
n
g
d
o
m
ain
an
alysis
and m
odel
e
d a
s
feat
u
r
es i
n
a
feat
ure
m
odel
.
In
DS
ky
L fea
t
ure m
odel is represe
n
ted as
a feature tree
whe
r
e
no
des re
p
r
esen
t
feat
ures an
d
edge
s rep
r
ese
n
t
t
h
e “sel
ect
i
o
n” rel
a
t
i
ons
hi
p
s
am
ong feat
u
r
es. Fr
om
a feat
ure
m
odel
,
a speci
fi
c vari
at
i
o
n of
a pro
d
u
ct
can be de
ri
ve
d by
sel
ect
i
ng t
h
e d
e
si
red feat
ures
base
d o
n
cust
o
m
er’
s
requirem
ents and
feature rela
tions
hips
ca
n
be specified in the feature model.
W
e
ca
n
have a
n
y
n
u
m
b
er
o
f
co
nfigu
r
ation
files. Bu
t o
n
l
y
o
n
e
will b
e
the cu
rren
t.
With
th
e im
p
l
e
m
e
n
tatio
n
o
f
t
h
e d
o
m
ain
th
ere are two
v
a
rian
ts: Fro
m
to
p to
bo
tto
m
o
r
fro
m
th
e bo
t
t
o
m
to
th
e to
p. Top
d
o
wn
ap
pro
a
ch
starts wi
th
th
e con
cep
t
in
the
br
oa
dest
sense
whi
c
h t
h
en i
n
c
l
udes (c
o
n
t
a
i
n
s
)
ot
he
rs. I
n
t
h
e
seco
nd case,
w
e
st
art
wi
t
h
t
h
e basi
c conce
p
t
s
.
W
e
have
used t
h
e t
op
do
w
n
ap
pr
o
ach i
n
DS
ky
L.
So we be
gi
n
wi
t
h
t
h
e co
nce
p
t
of M
o
o
d
l
e
appl
i
cat
i
o
n. To
creat
e
an
app
licatio
n
,
feature m
o
du
l
e
s are co
m
p
o
s
ed
t
o
g
e
t
h
er.
Th
e
file con
t
ainin
g
th
e class
defin
itio
n we
pu
t in
t
o
the folde
r
Clas
ses/Dom
a
in/Model/ a
n
d
re
na
m
e
it accordi
n
g to the last
pa
rt
of the
class
nam
e
: Moodle.
php.
A new
feat
ure is
created with the
fo
llowing
attrib
u
t
es and
then
th
e bu
si
n
e
ss lo
g
i
c is add
e
d
to
it.
public
Feat
ure
(
FeatureM
odel featureM
odel, String nam
e
)
{
this.featureModel
=
featureModel;
th
is.n
am
e
=
n
a
me;
th
is.m
an
d
a
to
ry
=
false;
th
is.con
cret
=
t
r
u
e
;
th
is.m
u
ltip
le
=
false;
th
is.h
i
d
d
e
n
=
false;
this.constrai
n
tSelected
=
fals
e;
t
h
i
s
.col
or =
ne
w c
o
l
o
r (t
hi
s);
t
h
i
s
.l
ocat
i
o
n =
new
feat
ureP
oi
nt
(
0
,
0
)
;
th
is.d
escrip
tion
=
nu
ll;
th
is.p
aren
t
=
nu
ll;
}
Out
t
ool
s
u
pp
o
r
t
s
di
f
f
e
r
ent
t
y
pes
of
rel
a
t
i
o
n
s
hi
p
bet
w
een
t
h
ese
feat
u
r
es
whi
c
h act
s as
ou
r
dom
ai
n o
b
j
ect
s.
A
few
o
f
t
h
em
al
on
g
wi
t
h
t
h
ei
r
not
at
i
o
ns a
r
e
r
e
prese
n
t
e
d
bel
o
w:
Op
tion
a
l
M
a
ndat
o
ry
Altern
ativ
e (OR)
R
e
qui
re
s
Uses
Im
pl
em
ent
a
ti
on
Gen
e
ralizatio
n
Co
m
p
o
s
itio
n
Mu
ltip
licities
are also
add
e
d to
th
efeatu
r
e
m
o
d
e
l. W
e
use th
e
m
u
ltip
lic
ity n
o
t
atio
n
to
ex
press th
e
constraint for t
h
e selection
of feature
s
from
the set. Th
e m
u
ltiplicity 1...* tells us, we
ha
ve to c
h
oose at
least
o
n
e
o
f
t
h
e possib
ilities, b
u
t
we mig
h
t
also
choo
se all o
f
th
em
. Th
ere are ru
les
wh
i
c
h
are asso
ciated
to
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
70
8 – 7
1
6
71
1
featu
r
es. Durin
g
t
h
e v
a
lid
ati
o
n
o
f
do
m
a
in
o
b
j
ects th
e
busin
ess log
i
c often
lo
ok
s
for th
e prop
erties o
f
the
objects present
.
This com
e
s
into ef
fect
du
ri
ng feat
ure sel
e
ct
i
on. S
o
m
e
ru
l
e
s whi
c
h are
prese
n
t
i
n
ou
r m
odel
are:
"X AL
WA
YS Y:
= If X
is
selected
the
n
Y
is
selected in eve
r
y valid confi
g
uration."
"X MAYBE
Y: = If X is selected
th
en
Y is
selected
in
at l
east on
e
b
u
t
not all v
a
lid
co
nfig
uration
s
.”
"X
NEVER
Y: = If X is selected
th
en
Y can
n
o
t
b
e
selected
in an
y
v
a
lid
co
nfigu
r
ation."
As a fi
rst
st
ep
feat
ures a
nd t
h
ei
r rel
a
t
i
o
nsh
i
ps are i
d
ent
i
f
i
e
d by
u
nde
rst
a
ndi
ng a
nd a
n
al
y
z
i
ng t
h
e
dom
ain. T
h
en
feature
s
are
re
prese
n
ted using the
feat
ur
e
m
odels. Based on the
feat
ure
s
which a
r
e se
lected,
specific program
s
are co
m
pose
d
. E
ach fe
ature will be im
plem
ented
as
a differe
nt
m
odule to enha
nc
e
reu
s
ab
ility. Thu
s
v
a
rian
ts of
an
ap
p
licatio
n are creat
ed
by selectin
g
and
im
p
l
e
m
en
tin
g
feat
u
r
es as
p
e
r t
h
e
req
u
irem
ents. In Fig
u
re
1
b
,
we depi
ct a si
m
p
le exa
m
ple of three clas
s
e
s and
four fe
ature m
o
dules. The
classes are ass
o
ciated to diffe
r
ent
feat
ur
es. I
n
or
der
t
o
gene
rat
e
a
p
r
o
g
r
am
,
classes of the
selected features are
com
posed
wi
t
h
a t
o
ol
. T
h
i
s
w
a
y
,
m
a
ny
di
ffer
e
nt
program
s
can
be c
r
eated
from
a set of fe
atures.
Havi
ng the
application t
o
pol
ogy
rea
d
y, the
ne
xt
step is t
o
m
odel depl
oym
e
nt topol
ogy and build
pl
an as speci
fi
ed i
n
Fi
gu
re 1
and
2. DS
ky
L uses B
P
M
N
n
o
t
a
t
i
on t
o
m
odel
t
h
e bui
l
d
pl
a
n
. F
o
r de
pl
oy
m
e
nt
, a
B
P
M
N
wo
rk
fl
ow
t
h
at
pr
o
v
i
s
i
ons
t
h
e
M
o
odl
e ap
pl
i
cat
i
on
o
n
Am
azon EC
2
vi
rt
ual
m
achi
n
es i
s
at
t
ache
d
.
Th
e
workflow i
n
stalls th
e app
licatio
n
s
as
d
e
fin
e
d in
th
e t
o
p
o
l
o
gy
an
d est
a
bl
i
s
hes t
h
e c
o
n
n
ect
sT
o rel
a
t
i
on
by
assi
gni
ng
t
h
e
I
P
ad
dr
ess
of
t
h
e M
y
SQL in
stan
ce to
t
h
e M
o
od
le co
nfig
ur
at
i
on on
t
h
e A
p
ache
W
e
b Se
rve
r
.
After
fin
i
sh
ing
m
o
d
e
ling
,
th
e b
ack
end
allows for expo
r
tin
g
a C
S
AR
file co
n
t
ai
n
i
ng
all requ
ired
d
e
fin
itio
ns.
Th
e
resu
ltin
g
CSAR
file d
e
p
l
o
y
s t
h
e im
p
l
e
m
en
tatio
n
art
i
facts and
t
h
e m
a
n
a
g
e
m
e
n
t
p
l
an
s to
app
r
op
riate
runtim
e environm
ents. Finally, the
users ca
n instantiate an application i
n
s
t
ance.
Fi
gu
re
1.
De
pl
oy
m
e
nt
M
odel
fo
r M
o
o
d
l
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Mo
del
Dri
v
e
n
Fr
a
m
ew
ork
f
o
r P
o
rt
abl
e C
l
ou
d
Servi
ces
(
A
pa
rna
Vija
ya)
71
2
Fi
gu
re
2.
B
u
i
l
d
Pl
an
f
o
r M
o
od
l
e
The Cloud Se
rvice Archi
v
e
[12],
[13]
is a co
n
t
ain
e
r
file wh
ich
ho
ld
s t
h
e
serv
ice tem
p
late o
f
a cloud
application, all
artifacts
requi
r
ed to
m
a
nage
the lifecycle
of the c
o
rres
p
on
di
n
g
cl
o
u
d
appl
i
cat
i
o
n (i
.
e
. t
h
e
im
pl
em
ent
a
t
i
o
n a
r
t
i
f
act
s o
f
t
h
e
ope
rat
i
o
ns
of
t
h
e
n
ode
typ
e
s) as
well as all artifacts t
o
ex
ecu
te th
e
clo
ud
application (i.e
. the de
ploym
e
nt artifacts of the node
typ
e
s lik
e v
i
rt
u
a
l i
m
a
g
es, EJBs,
WSDL files, SQL
DDL
etc). These fil
e
s are typically orga
ni
zed in several subdirectories each
of
whic
h contains related files
.
Each
C
S
AR
m
u
st
co
nt
ai
n a su
bdi
re
ct
ory
cal
l
e
d M
e
t
a
-In
f. T
h
i
s
subdirect
ory must contain a
so
-called
m
a
n
i
fest file.
Th
is file is n
a
med
MANIFEST and
h
a
s th
e file ex
ten
s
io
n
.MF. It represen
ts m
e
tad
a
ta o
f
th
e o
t
h
e
r files in
th
e
CSAR. T
h
ese
metadata are give
n in the form
at of
nam
e
/
v
al
ue p
a
i
r
s.
These
nam
e
/
v
al
ue pai
r
s
(
<name
>
:
<val
ue>
) are
or
ga
ni
zed i
n
bl
ock
s
. Eac
h
bl
o
c
k p
r
o
v
i
d
es m
e
t
a
data of a ce
rtain artifact of the CSAR. T
h
e first
b
l
o
c
k
o
f
th
e m
a
n
i
fest
file prov
id
es m
e
tad
a
ta
of th
e CS
AR
itself (e.g
. its v
e
rsion
,
creator etc) as
fo
llows:
M
a
ni
fest
-
V
ersi
on:
x.
x
C
S
AR
-
V
ersi
on
:
x.y
Created-By: te
st
En
try-Serv
ice-Tem
p
late
: file n
a
m
e
th
at is th
e en
try
po
in
t
for th
e clou
d applicatio
n
.
Each
o
t
h
e
r
b
l
ock
b
e
g
i
ns with
a n
a
m
e
/v
alu
e
p
a
ir th
at po
in
t to
an
artifact wi
th
in
th
e CSAR
b
y
m
ean
s o
f
a p
a
th-
nam
e
as sho
w
n
i
n
Fi
gu
re
3.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
70
8 – 7
1
6
71
3
Fi
gu
re 3.
M
a
ni
fest
file
The st
r
u
ct
u
r
e
of a
ppl
i
cat
i
o
ns
i
s
defi
ne
d by
a t
o
p
o
l
o
gy
a gra
p
h o
f
t
y
ped
no
des a
nd
di
r
ect
ed t
y
ped
edge
s. N
o
des
rep
r
ese
n
t
com
p
o
n
e
n
t
s
f
o
rm
ing a
n
ap
pl
i
cat
i
on /
feat
u
r
es
and e
d
ges de
fi
ne t
h
e rel
a
t
i
o
n
s
and
depe
n
d
enci
es
bet
w
ee
n t
h
em
. For i
n
st
ance,
t
h
e t
o
p
o
l
o
gy
o
f
t
h
e m
oodl
e a
ppl
i
cat
i
o
n (Fi
g
ure
1) c
o
n
s
i
s
t
s
of t
h
e
act
ual
P
H
P m
o
d
u
l
e
,
an
A
p
a
c
he
We
b Se
r
v
er, a
M
y
SQL
dat
a
base
, t
w
o
ope
rat
i
n
g sy
st
e
m
s (one
fo
r t
h
e
W
e
b
serve
r
an
d
one
fo
r t
h
e M
y
SQ
L dat
a
base
), a
n
d t
w
o
vi
rt
ua
l
m
achi
n
es. T
h
e
rel
a
t
i
ons
hi
ps i
n
t
h
i
s
t
o
pol
ogy
defi
ne
,
fo
r i
n
st
a
n
ce, t
h
at
t
h
e M
o
odl
e
appl
i
cat
i
o
n i
s
host
e
d
on
a
Web se
rve
r
a
nd t
h
at
t
h
e a
ppl
i
c
a
t
i
on c
o
n
n
ect
s t
o
t
h
e
M
y
SQL dat
a
b
a
se. The
t
y
pes
of
n
odes a
n
d r
e
l
a
t
i
onshi
ps s
p
eci
fy
t
h
ei
r p
r
o
p
ert
i
e
s a
nd m
a
nagem
e
nt
o
p
er
at
i
ons.
The t
y
pe
A
p
ac
he
W
e
b
Ser
v
er
defi
nes
pr
ope
rt
i
e
s, suc
h
as
p
o
r
t
o
r
ve
rsi
o
n,
a
nd m
a
nagem
e
nt
ope
rat
i
o
n
s
, s
u
ch as
st
art
or
depl
oy
.
The act
ual
i
m
pl
em
ent
a
t
i
on o
f
a n
ode i
s
pr
o
v
i
d
e
d
by
one
o
r
m
a
ny
Depl
oy
m
e
nt
Art
i
f
act
s,
e. g.
,
a Linux VM i
m
age, an ope
r
ating sy
stem
package
for the
Ap
ache
W
e
b
Server, or a
n
a
r
chi
v
e c
ontaini
ng the
PHP files of Mo
od
le. In
add
itio
n, typ
e
s
m
a
y
d
e
fi
n
e
Im
p
l
e
m
en
tatio
n
Artifacts th
at i
m
p
l
e
m
en
t th
e
m
a
n
a
g
e
m
e
n
t
o
p
e
ration
s
for th
e resp
ectiv
e
ele
m
en
t. Th
e ap
p
lication
to
polo
g
y
and
related
artifacts are b
und
led
in
to
a Clo
ud
Service ARchi
v
e (CSAR),
which is a
st
and
a
rdi
z
e
d
pac
k
ag
i
ng f
o
rm
at
for appl
i
cat
i
o
ns. A
ppl
i
cat
i
o
n t
o
p
o
l
ogi
es
can b
e
p
r
oc
ess
e
d i
n
a
n
i
m
perat
i
v
e or
declarative way: Im
perative proces
sin
g
relies on
the i
m
p
l
e
m
en
tati
o
n
of
m
a
nagem
e
nt
p
l
ans t
h
at
ca
n
b
e
exec
ut
ed
f
u
l
l
y
aut
o
m
a
t
e
d t
o
per
f
o
r
m
t
h
e desi
re
d m
a
nag
e
m
e
nt
t
a
sk, e
.
g.,
t
o
in
stan
tiate, b
a
ck
up
,
up
grade, o
r
te
rm
inate an application. T
h
ese
hi
gh le
vel m
a
n
a
gem
e
nt tasks are
im
pl
em
ent
e
d by
orche
s
t
r
at
i
n
g
l
o
w l
e
vel
m
a
nagem
e
nt
ope
r
a
t
i
ons p
r
ovi
de
d by
Im
pl
em
ent
a
t
i
on
Art
i
f
ac
t
s
of
no
des a
n
d rel
a
t
i
ons
hi
ps
. Si
nce
t
h
e m
a
nagem
e
nt
pl
a
n
s are
t
y
pi
cal
l
y
im
pl
ement
e
d
by
t
h
e a
p
pl
i
cat
i
on
devel
ope
r,
t
h
ey
enabl
e
op
erat
or
s t
o
m
a
nage t
h
e a
ppl
i
c
a
t
i
on by
r
u
nni
n
g
p
r
ede
f
i
n
e
d
p
l
ans wi
t
h
out
t
h
e need t
o
u
n
d
e
r
st
an
d
al
l
t
h
e t
echni
cal
det
a
i
l
s
of
t
h
e m
a
nagem
e
nt
t
a
sk. Te
chni
cal
l
y
, m
a
nagem
e
nt
pl
a
n
s are
i
m
pl
em
ent
e
d as
wo
rk
fl
o
w
s.
De
cl
arat
i
v
e p
r
oce
ssi
ng
,
on
t
h
e
ot
he
r
han
d
,
sh
i
f
t
s
t
h
e
depl
oy
m
e
nt
and
m
a
nagem
e
nt
l
ogi
c
fr
om
pl
ans
t
o
t
h
e r
u
nt
im
e. To
per
f
o
rm
t
h
e af
ore
m
ent
i
oned
hi
g
h
-l
e
v
el
m
a
nagem
e
nt
t
a
sks, t
h
e r
unt
i
m
e has t
o
kn
o
w
th
e op
eratio
ns
th
at h
a
v
e
t
o
b
e
called
an
d th
eir ord
e
r.
Decl
a
r
at
i
v
e pr
ocessi
ng
i
s
wel
l
sui
t
e
d fo
r
t
h
e depl
o
y
m
e
nt
o
f
sim
p
le ap
p
licatio
n
s
bu
t is no
t ab
le to
facilitate co
m
p
lex
man
a
g
e
m
e
n
t
ta
sk
s fo
r
v
a
ri
o
u
s k
i
nd
s of app
licatio
n
structures.
In
o
r
de
r t
o
po
rt
an a
p
pl
i
cat
i
on,
•
Speci
fy
t
h
e
no
des a
n
d
no
de t
y
pes a
n
d
m
odel
t
h
e ap
pl
i
cat
i
o
n se
rvi
ce t
o
p
o
l
ogy
.
•
Create artifacts
of the
softwa
re
wh
ich
n
e
ed
s to
b
e
in
stalled
.
•
Ide
n
t
i
f
y
an
d
de
scri
be t
h
e
rel
a
t
i
ons
hi
p
bet
w
ee
n t
h
e
com
p
o
n
e
n
t
s
.
•
Gen
e
rate th
e bu
ild
p
l
an
and
write th
e resp
ectiv
e p
a
ram
e
te
r to inv
o
k
e
t
h
e
serv
ice.
•
Upl
o
ad it to t
h
e container and invoke
th
e serv
ice to
u
s
e th
e
ap
p
lication
.
Po
rtab
ility is en
sured
b
y
th
e t
w
o eng
i
n
e
s
work
i
n
g tog
e
th
er
wh
en
b
i
n
d
i
n
g
man
a
g
e
m
e
n
t
plan
s.
Strict
separation of
architectural com
pone
nts through wellde
f
ined
OS
Gi interfaces e
n
able
s the replace
m
e
nt of
im
plem
entatio
ns of c
o
m
ponents. T
h
is also allows
each com
pone
nt to be
scaled i
nde
pe
nde
ntly.
Afte
r
upl
oadi
ng
t
h
e
C
S
AR
, t
h
e
dep
l
oym
ent
of
t
h
e
appl
i
cat
i
o
n f
o
l
l
o
ws
t
h
ree st
e
p
s
:
1.
Fi
rst
,
t
h
e C
S
A
R
i
s
unpac
k
ed
and t
h
e fi
l
e
s ar
e put
i
n
t
o
t
h
e F
i
l
e
s st
ore, whi
c
h i
s
backe
d
ei
t
h
er by
t
h
e l
o
ca
l
fi
l
e
sy
st
em
or
Am
azon S
3
.
2.
The
n
, t
h
e
XM
L fi
l
e
s are
l
o
a
d
ed a
n
d
p
r
ocess
e
d t
o
de
pl
oy
t
h
e rel
a
t
e
d
de
pen
d
enci
es
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Mo
del
Dri
v
e
n
Fr
a
m
ew
ork
f
o
r P
o
rt
abl
e C
l
ou
d
Servi
ces
(
A
pa
rna
Vija
ya)
71
4
3
.
Fin
a
lly,
th
e p
o
rtab
ilityAPI read
s
th
e
top
o
l
og
y
m
o
d
e
l,
in
st
an
ce d
a
ta, p
r
op
erties o
f
no
d
e
s
lik
e
th
e po
rt o
f
web server
etc and relationshi
ps
to
provision and confi
g
ur
e
the Cloud a
ppl
ication accordi
ng t
o
the
build
pl
an.
The
depl
oy
ed
appl
i
cat
i
o
n can be
i
n
stanti
ated by exec
uting the
bu
ild p
l
an
o
f
the ap
p
lication.
Cr
ed
en
tials (
e
.
g
.
, fo
r
A
m
azo
n EC2
)
or
co
nf
i
g
ur
ation
s
(
e
.
g., m
achine size) are
passe
d as i
n
put m
e
ssage to the
wo
rk
fl
o
w
. T
o
i
n
st
ant
i
a
t
e
M
oodl
e, t
h
e b
u
i
l
d
pl
an fi
r
s
t
st
art
s
t
w
o vi
rt
ual
m
achi
n
es wi
t
h
a Li
nu
x o
p
er
at
i
n
g
sy
st
em
and i
n
st
al
l
s
Apache
Web
Ser
v
er a
nd M
y
S
Q
L
on
them
. Then, i
t
uses the re
s
p
ective m
a
nage
m
e
nt
o
p
e
ration
s
to
in
stall th
e PHP ap
p
lication
,
imp
o
rt th
e d
a
t
a
ba
se schem
a
, and est
a
bl
i
s
h t
h
e d
a
t
a
base co
nnec
t
i
o
n
.
Aft
e
r c
o
m
p
l
e
t
i
on
, a bui
l
d
pl
a
n
m
a
y
ret
u
rn c
e
rt
ai
n i
n
f
o
rm
ati
on, f
o
r exam
pl
e, t
h
e web ad
dress
of t
h
e de
pl
oy
e
d
application ins
t
ance. T
h
e Moodle
build
pl
a
n
ret
u
rns t
h
e
U
R
L of t
h
e r
u
nn
i
ng M
o
o
d
l
e
i
n
s
t
ance,
whi
c
h i
n
cl
ude
s
th
e pub
lic U
R
L of
t
h
e
v
i
r
t
u
a
l
m
ach
in
e
r
unnin
g
t
h
e
Ap
ache
W
e
b
Serv
er.
Th
e scr
e
en
shots o
f
im
p
l
e
m
en
tatio
n
fo
r
po
rt
i
n
g a
p
p
l
i
cat
i
on f
r
om
HP
H
o
ri
zo
n
an
d
Ope
n
St
ack t
o
Am
azon EC
2 a
r
e s
h
o
w
n i
n
t
h
e f
o
l
l
o
wi
n
g
f
i
gu
res.
Fi
gu
re 4.
HP Hel
i
o
n
I
n
st
anc
e
F
i
g
u
r
e
5
.
O
p
en
S
t
a
c
k
in
s
t
an
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
70
8 – 7
1
6
71
5
Fi
gu
re 6.
Am
azon
W
e
b Ser
v
i
ce
i
n
st
ance
3.
PR
ELIMINAR
Y EVA
L
UATION
We
have
p
o
rt
e
d
t
h
e M
o
odl
e
Ap
pl
i
cat
i
on
fr
om
t
w
o di
f
f
ere
n
t
cl
o
u
d
en
vi
r
onm
ent
s
O
p
en
st
ack an
d
H
P
Ho
ri
zo
n C
l
ou
d t
o
Am
azon
A
W
S a
n
d
res
u
l
t
s
i
m
pl
y
t
h
at
t
h
e a
ppl
i
cat
i
o
n
i
s
de
pl
oy
e
d
o
n
Am
azon A
W
S
success
f
ully. T
h
e tables
bel
o
w s
h
owcase
the tim
e
differe
nce and
platform
used for port
ing.
Tab
l
e 2
.
Tier Nod
e
Defin
ition
Tem
p
late
Server Prope
r
ty
OpenS
t
ac
k
HP
Hori
z
o
n
Nu
m
b
er
of CPUs
4
2
M
e
m
o
ry
Size
8GB
4GB
Disk Size
80GB
50GB
Initial instances
1
1
Secur
ity Rule Pr
otocols
HT
T
P
HT
T
P
Secur
ity Rule Por
t
s
22,
80,
44
3
22,
80,
44
3
Tabl
e 3.
Ti
m
e
Eval
uat
i
o
n f
o
r po
rt
i
n
g
M
o
o
d
l
e
Sour
ce Server
T
r
ansfer
Dir
ectio
n
T
i
m
e
HP
Hor
i
zon
Am
azon E
C
2
331sec
Openstack Am
azon
E
C
2
327sec
We ha
ve al
so
t
r
i
e
d t
o
est
i
m
a
t
e
t
h
e effo
rt
re
qui
red f
o
r p
o
rt
i
ng t
h
e ap
pl
i
c
a
t
i
on t
o
cl
o
ud
u
s
i
ng C
l
o
ud
M
i
grat
i
on P
o
i
n
t
(C
M
P
) [1
4]
. The fi
nal
val
u
e
of C
M
P i
s
det
e
rm
i
n
ed as a w
e
i
ght
ed s
u
m
of i
t
s
four com
p
o
n
ent
s
CMP
i
with i
∈
{co
n
n
ection, co
d
e
ch
ang
e
s, in
stallatio
n an
d co
nfigu
r
ati
o
n, d
a
tab
a
se}. Th
e
co
m
p
lex
ity
lev
e
l
for co
nn
ections th
at
will b
e
affected
d
u
ring
p
o
rting
is
LOW si
n
ce liv
em
i
g
ration
is no
t co
n
s
i
d
ered in
t
h
e scope
o
f
t
h
e
wo
rk
.
It
is con
s
id
ered
th
at an
app
licatio
n
is
po
r
t
ed fr
o
m
LA
N to
LA
N.
Th
e c
o
de
changes
are m
a
inly
applicable i
n
classes that are
re
sp
o
n
si
bl
e f
o
r
com
m
uni
cat
i
on bet
w
een e
x
t
e
rnal
sy
st
em
s.
Que
r
y
m
odi
fi
cat
i
on
tasks associated to
database c
h
anges
will not be accounta
b
le here since t
h
e sam
e
database will be ported t
o
th
e tag
e
t clou
d. Tasks related to
in
stallatio
n an
d
con
f
i
g
u
r
a
t
i
on o
f
t
h
i
r
d
p
a
rt
y
l
i
b
rary
an
d dat
a
base
doe
s n
o
t
require a
n
y m
a
nual e
f
fort but
changes
nee
d
t
o
be
done
i
n
the build
plan and
de
pl
oym
e
nt m
odel accordingly.
Fo
r th
e Mood
le ap
p
lication
po
rtab
ility we h
a
v
e
id
en
tifie
d
3
6
Cl
o
u
d
Migratio
n
Po
in
ts and
th
e effo
rt in
h
ours
is less th
an
3
ho
urs.
4.
CO
NCL
USI
O
N
Th
is
p
a
p
e
r
p
r
esen
ted
DSk
y
L, a to
o
l
d
e
v
e
loped
for app
licatio
n
po
rtab
ility acro
s
s m
u
ltip
l
e
clo
u
d
s
b
y
usi
n
g m
odel
d
r
i
v
en a
r
chi
t
ect
u
r
e.
We
descri
b
e
d t
h
e
p
r
o
b
l
e
m
ha
ndl
e
d
by
t
h
e t
ool
i
n
sect
i
o
n
1 an
d s
u
m
m
a
ri
zed
the m
e
thod
be
hind t
h
e tool a
n
d its m
a
in functionalities with
its e
v
aluations i
n
sections
2 a
n
d
3.
Our a
p
proach
w
a
s tested
w
ith
thr
ee cloud
v
e
ndo
r
s
. Our
g
o
a
ls
f
o
r
f
u
t
u
r
e
wo
rk
i
n
clud
e th
e im
p
r
ove
m
e
n
t
o
f
D
S
kyl b
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Mo
del
Dri
v
e
n
Fr
a
m
ew
ork
f
o
r P
o
rt
abl
e C
l
ou
d
Servi
ces
(
A
pa
rna
Vija
ya)
71
6
su
ppo
rting
p
r
ov
ision
fo
r d
a
t
a
po
rtab
ility. Fu
rt
h
e
r wo
rk
sh
ou
l
d
also ad
dress m
o
re co
m
p
lex
app
licatio
n
s
sup
p
o
rt
i
n
g a
d
d
i
t
i
on an
d
rem
oval
o
f
feat
ures
and
cl
o
ud
ser
v
ices on
the go
in
o
r
d
e
r to
en
su
re t
h
at realistic
i
n
st
ances
o
f
re
qui
rem
e
nt
s are bei
n
g
use
d
,
t
h
ro
u
g
h
di
rect
c
o
nt
act
wi
t
h
t
h
e
busi
n
ess
an
d c
u
st
om
ers.
REFERE
NC
ES
[1]
Gurudatt Kulkar
ni Cloud Compu
ting Software as
Service,
Interna
tional Journal of Cloud Computing and Services
Scien
c
e
(
I
J-CL
OSER)
,
Vol 1
,
201
2 pages 11
-16.
[2]
P
r
ati
y
us
h Guleri
a Guleri
a, Vikas
S
h
ar
m
a
, M
a
nis
h
Arora, Develo
pm
ent and Us
age of S
o
ftware as
a S
e
rvice for
a
Cloud and Non
C
loud Based
Envir
onment An
Empirical Stud
y,
International
Journal of Cloud
Computing and
Services Science (
I
J-CLOSER)
,Vol 2 No 1
,
201
3 pages 50
-58.
[3]
Aparna Vij
a
ya
,
Neel
anara
y
a
n
a
n
V,
“Framework for Pla
tfor
m
Agnostic
En
terprise Applicat
ion Development
Supporting Multiple C
l
ouds”
, Proc. S
y
mp of BigData and C
l
oud compu
ting Challeng
es - Elsevier Proced
ia
Computer Scien
ce
, 2015
, Volu
me 50, pp
73-80.
[4]
Aparna Vij
a
y
a
,
Pritam
Dash, Neelan
aray
anan V,
“Migration
of L
e
gacy
Ent
e
rprise App
lications to
Multiple Clouds:
A F
eature b
a
s
e
d appro
ach”
.
Lecture No
tes on Software Engineering
(LNSE, ISSN:
2301-3559, DOI:
10.7763/LNSE )
Journal.
[5]
N. Loutas, E. K
a
materi
, and K. Taraban
i
s,
―
A
Semantic Inter
operabilit
y Fram
ework for Cloud Platform as
a
Service
, I
E
EE Third Intern
ational Conferen
ce on
Cloud Com
puting Techno
log
y
an
d Science (Clou
d
Com), Athens
,
2011, pp
. 280–2
87.
[6]
Fotis Gonidis,
Iraklis Par
a
skak
is, Dim
itrios K
ourtesis,
Addressing the Challenge of
Application Portability in
Cloud Pla
tforms
, Ba
lkan
Confer
e
n
ce
in Inform
at
i
c
s
,
BCI
'13
,
Greece, Sep
t
ember,
2013.
[7]
D. P
e
tcu, G. M
acariu
, S
.
P
a
nica
,
and C. Cr
ă
ciun,
―
P
o
rtable Clo
ud applic
ations
—from
theor
y
t
o
practi
ce,
Fu
tu
re
Generation Com
puter Systems
, 2
012.
[8]
Magdalen
a
Kostoska, Marjan G
u
se
v, Sasko Ris
t
ov, A New Cloud Se
rvices Portabil
ity
Platfor
m
,
24th DAAAM
International
Symposium on Intellig
en
t Manu
fact
uring and Au
tomation
, 2013.
[9]
Danilo Ardagn
a
,
E
lisabe
tta
Di,
Giuliano C
a
sal
e
, Dana
Petcu
,
Parastoo Mohag
h
eghi
, S´e
b
a
s
tien Mosse
r,
Pe
te
r
M
a
tthews
,
Anke Gerick
e, C
y
ri
l Ball
agn
y
, F
r
a
n
ces
co
D’Andria, Cos
m
in-S
epti
m
i
u Nechifor,
Craig S
h
eridan
,
MODACLOUDS: A Model-Dri
v
en Approac
h fo
r the Design an
d Exe
c
ution o
f
Applica
tions on
Multipl
e
Cloud
s
,
MiSE-2012.
[10]
Redhat: http://www.
redhat.com/develop
ers/open
s
hift/ (2015)
.
[11]
Fotis Gonidis, Iraklis Paraskakis,
Anthon
y
J.
H.
Simons,
Di
mitrios Kourte
sis,
Cloud
Applicat
ion
Portabilit
y: An
Initial
V
i
ew,
Balkan Confer
ence
in Informatics, B
C
I '
1
3
,
Thessalo
niki,
Gree
ce, September, 2013.
[12]
Tobias Binz, Uwe Breitenb
¨
u
ch
er, Florian Hau
p
t, Oliv
er Kopp, Frank Ley
m
an
n, "TOSCA: Po
rtable Automated
Deplo
y
men
t
and
Management of
Cl
oud Applications". http:
//www.iaas.un
i
-stuttg
art.
d
e
/RUS-data/INBOOK-2014
[13]
Gerd Breiter
,
Frank Ley
m
ann, Thomas Spatzier,
" Topol
og
y
and
Orchestration Specif
i
cationfor Cloud Application
s
(TOSCA):Cloud Service Arch
ive (CSAR) Version 0.1", 2012.
[14]
Van T.K. Tr
an,
Kevin Lee, Alan
Fekete, Anna Liu,Jack
y
Keung,
Size Estimation
of Cloud Mi
gration Projects with
Cloud Migratio
n Point (
C
MP)
, Intern
ation
a
l S
y
m
posium
on
E
m
pirical
So
ftware Eng
i
neer
ing
and Measurement,
2011.
Evaluation Warning : The document was created with Spire.PDF for Python.