Tempo al-Awa eness in SLAs: Why Should We
Be Conce ned?
C. Mülle , A. Ruiz-Co és, and P. Fe nández
Dp o. Lenguajes y Sis emas In o má icos
ETS. Ingenie ía In o má ica - Uni e sidad de Se illa (Spain - España)
41012 Se illa (Spain - España)
{cmulle ,a uiz,pablo m}@us.es
Abs ac . T adi ionally, Se ice Le el Ag eemen s ha e been decom-
posed in wo se s o p ope ies: unc ionals (wha ) and non- unc ionals
(how). Howe e , in ou opinion, he e has been a hi d key elemen ha
has had a mino a en ion om academy: empo al awa eness (when).
We belie e empo ali y is a main conce n ha should be add essed in
ealis ic scena ios. In doing so, his posi ion pape discuss ou expe ience
in ex ending he speci ica ion WS-Ag eemen wi h a empo al Domain
Speci ic Language; impo an ly, main aim o he pape is o p o oke a
deba e abou he impo ance o empo ali y in SLAs.
1 In oduc ion
As se ice-o ien ed compu ing (SOC) [18] pa adigm is being widely adop ed
in indus y, new challenges need o be conside ed o add ess he p oblems o
complex scena ios. The co e in as uc u e s ack behind SOC has been p o ed
as solid pila o a icula e new specifica ions o deal wi h mo e abs ac business
equi emen s. In pa icula , a key elemen is ha gaining impo ance is he
adi ional se ice le el ag eemen (SLA) idea o en ich he basic SOA pla o m.
In doing so, SLAs will help o exp ess and gua an ee he app op ia e quali y o
se ice (QoS) demanded by he diffe en pa icipan s o a SOC scena io.
Howe e , in spi e o he ex ensi e wo k ca ied ou on SLAs in he con ex
o SOC [7,8,15,16], we belie e he e has been no needed mino a en ion o an
impo an issue ha mus be p omo ed as a fi s -class elemen o SLAs: he
empo ali y.
T adi ionally, SLAs a e exp essed using a se o p ope ies in o de o spec-
i y “which” se ice is offe ed and “how” i is offe ed. Tha is o say, i includes
equi emen s and gua an ees abou unc ional (FP), and non- unc ional p ope -
ies (NFP) o he se ices. Howe e , ano he impo an ques ion abou se ices
is “when”. In ou opinion, empo al awa eness is op conce n in o de o deal
wi h ealis ic scena ios because hey usually ha e s ong empo al es ic ions
a diffe en le els ha should be me : he en i e ag eemen (e.g. he ag eemen
This wo k has been pa ially suppo ed by he Eu opean Commission (FEDER) and
Spanish Go e nmen unde CICYT p ojec Web-Fac o ies (TIN2006-00472).
expi es on 2007/05/31), any FP o he se ice (e.g. his ope a ion o he se ice
is a ailable om 8:00h o 18:00h), and/o any NFP ha appea s in he SLA
(e.g. he esponse ime is 30 ms om 8:00h o 17:00h and 15 ms om 17:00h o
8:00h).
In o de o deal wi h he p e ious issues, in [13] we p opose a amewo k o
exp ess empo al-awa e SLA by ex ending he specifica ion WS-Ag eemen (WS-
Ag) [7]. This specifica ion is a p oposed ecommenda ion o he Open G id Fo um
wo king g oup (OGF) ha p o ides bo h a s uc u al schema o speci ying
SLAs and a p o ocol o c ea ing hem based on a mechanism o empla es. I is
impo an o highligh ha , due o compa ibili y and complexi y easons, WS-Ag
specifica ion only defines a gene al s uc u e o he ag eemen ; complemen a y, in
o de o en ich he co e s uc u e, WS-Ag lea e ex ensibili y poin s o domain-
specific ex ensions like [1], [17] and [14].
Based on he expe ience o he au ho s in c ea ing he empo al-awa eness
ex ension o WS-Ag, he main goal o his posi ion pape is o analyse he
key opics o empo ali ies in SLAs: when i is use ul? o which is he co ec
app oach? a e some o he ques ions we aise. Addi ionally, wi h his discussion
we aim o boos a deba e o make empo ali y he main conce n o be add essed
by he SLA communi y.
The es o he pape is s uc u ed as ollowing: In sec ion 2, we ou line he
main issues o ou empo al-awa e DSL o ex ending Ws-Ag. Nex , in sec ion
3, we analyse om a c i ical poin o iew ou expe ience. In sec ion 4, we
summa ise a se o discussion opics based on ou expe ience. Finally, we p esen
he conclusions.
2 Imp o ing Tempo al-Awa eness o WS-Ag in a Nu shell
2.1 WS-Ag eemen in a Nu shell
WS-Ag specifies an XML-based language and a p o ocol o ad e ising he
capabili ies o se ice p o ide s, c ea ing ag eemen s based on ag eemen offe s.
The s uc u e o an ag eemen o WS-Ag is comp ised o :
–Name: iden ifies he ag eemen and can be used o e e ence.
–Con ex : i includes in o ma ion such as he name o he pa ies and hei
oles o ini ia o o esponde o he ag eemen . Addi ionally, i can e e o
an ag eemen empla e i needed. In his elemen , an ag eemen li e ime can
be defined by means o an elemen called “Expi a ionTime”.
–Te ms: ag eemen e ms a e w apped by e m composi o s, which allow
simple e ms o se s o e ms o be deno ed by “Exac lyOne”, “OneO Mo e”,
o “All”. The wo main ypes o e ms a e: (a) Se ice e ms: hey p o ide
in o ma ion o ins an ia e o iden i y se ices and ope a ions in ol ed in
he ag eemen . Addi ionally, i can comp ise o in o ma ion abou he mea-
su able se ice p ope ies. (b) Gua an ee e ms: hey desc ibe he se ice
le el objec i es (SLO) ag eed by he pa ies. I also includes he scope o
he e m (e.g. a ce ain ope a ion o he se ice o he whole se ice i sel );
a “Quali yingCondi ion” ha specifies he alidi y condi ions unde which
he e m is applied; and in o ma ion abou business p ope ies in he “Busi-
nessValueLis ” elemen such as “Impo ance”, “Penal y” o “Rewa d” and
“P e e ence”.
In o de o c ea e ag eemen s, WS-Ag allows us o speci y empla es wi h he
abo e s uc u e, bu including ag eemen “C ea ion Cons ain s” ha desc ibe
he a iabili y allowed by a pa y and i should be aken in o accoun du ing
he ag eemen c ea ion p ocess. The elemen “I em” is used in case we need o
desc ibe a c ea ion cons ain o a specific pa o he ag eemen .
2.2 Tempo al-Awa eness o WS-Ag eemen
WS-Ag iden ifies wo loca ions o include empo al awa eness. On he one hand,
li e ime o he en i e ag eemen mus be included in Con ex in o he “Expi a-
ionTime” elemen (i.e. he las ins an whe e he ag eemen is alid). On he
o he hand, WS-Ag ecommends he use o “Quali yingCondi ion” elemen s o
desc ibing alidi y pe iods o e ms and/o he pa y p e e ences. Howe e , he
specifica ion documen lea es open he specific way ha empo al awa eness
mus be exposed o easons o compa ibili y and complexi y.
2.3 Ou Tempo al DSL o WS-Ag eemen
Figu e 1 shows an UML class diag am which ep esen s he schema we p o-
posed in [13] o desc ibing empo al p ope ies in SLAs. An in e al is he basic
elemen , diffe en non-disjoin in e als can be g ouped oge he so ha mo e
ini : Da eTime
du a ion : loa
du a ionMe ic : {Seconds,
Hou s, Days,
Mon hs}
1
1..*
<<in e ace>>
In e al
<<in e ace>>
Disjoin
op : {And, O , Xo }
<<in e ace>>
Pe iodical
epe i ions : loa
equency: loa
equencyMe ic : {Seconds,
Hou s, Days,
Mon hs}
{incomple e}
1
1
Fig. 1. Schema o Tempo al In e als
complex in e als can be composed. The h ee in e aces deno e he diffe en
ypes o in e als:
–In e al: s ands o he basic elemen ; is comp ised o an ini ial ime and a
du a ion (which can be infini e) exp essed in seconds, hou s, days, o mon hs.
– Disjoin : s ands o disjoin in e als cons i u ed o a se o in e als ela ed
by a logic ope a o (o , and, o xo ).
–Pe iodical: s ands o pe iodic in e als, ei he disjoin o non-disjoin . I s
pe iodici y is comp ised o he numbe o pe iod epe i ions (which can be
infini e) and a equency exp essed in seconds, hou s, days, o mon hs, which
deno es he ime be ween wo consecu i e in e als.
Ag eemen Templa e
Te ms Composi o
Con ex
Name
<Se iceDesc ip ionTe m>
<Reques > </…>
<Algo i hm> </…>
<MTBR> </…>
</…>
<Se iceP ope ies>
…
</…>
<Se iceP o ide > esponde </…>
…
<All>
<All>
…
C ea ion Cons ain s
<I em Name="Tes P e ious">
<Loca ion>
//Se iceDesc ip ionTe m/MTBR
</…>
<I emCons ain >
< es ic ion base="posi i eIn ege ">
<minInclusi e alue="40"/>
</…>
</…>
<In e al ini ="2007-09-30T
00:00:00+01:00"
du a ion=“1.0"
du a ionMe ic="Days"/>
</…>
<I em Name=”Algo i hmAllowed”>
<Loca ion>
//Se iceDesc ip ionTe m/Algo i hm
</…>
<I emCons ain >
< es ic ion base="xs:s ing">
<enume a ion alue=“Dijks a" />
<enume a ion alue=“Knapsack" />
<enume a ion alue=“K uskal" />
</…>
</…>
</…>
<Cons ain >
<In e al ini ="2007-09-29T
00:00:00+01:00"
du a ion=“2.0"
du a ionMe ic="Days"/>
</…>
<Gua an eeTe m
Obliga ed=“Se iceP o ide ”…>
<Quali yingCondi ion>
</…>
<SLO>”MTBR ≥ 20”</…>
<BusinessValueLis >
<Cus omBussinessValue>
<Va iableRe e ence>Algo i hm</…>
<U ili yFunc ion>F1(U)</…>
<Va iableRe e ence>MTBR</…>
<U ili yFunc ion>F3(U)</…>
</…>
</…>
</…>
(a)
(b)
(c)
<GlobalPe iod>
<Pe iodical equency=”1.0”
equencyMe ic=”Hou s”
epe i ions=”13.0”>
<In e al ini =”2007-10-01T00:00:00+01:00”
du a ion=”167.0”
du a ionMe ic=”Hou s”/>
</…>
</…>
F om Monday 0:00
o Sunday 23:00
F om 10/01/2007
o 12/30/2007
(13 weeks)
<Pe iodical equency=”62.0”
equencyMe ic=”Hou s”
epe i ions=”13.0”>
<Pe iodical equency=”14.0”
equencyMe ic=”Hou s”
epe i ions=”5.0”>
<In e al ini =”2007-10-01T
08:00:00+01:00”
du a ion=”10.0”
du a ionMe ic=”Hou s”/>
</…>
</…>
F om 8:00
o 18:00
F om 18:00
o 8:00
F om Monday
o F iday
F om 18:00 o 24:00
+
Weekend
+
F om 0:00 o 8:00
Validi y
Pe iod o he
en i e ag eemen
Validi y Pe iod o :
(a) Se ice
Desc ip ion
Te ms
(b) The SLO
(c) The P e e ences
Validi y
Pe iod o a single
cons ain
(Tes P e ious)
Validi y
Pe iod o all
c ea ion
cons ain s
Fig. 2. An example o empo al-awa e WS-Ag Templa e
Ou p oposal allows o include empo ali y ega ding se e al aspec s o ag ee-
men s. Figu e 2, shows a pa icula case s udy whe e a p o ide offe s compu ing
se ices o o he o ganiza ions wi h a empla e. The Figu e includes whe e we
imp o e he Tempo al-Awa eness in o he WS-Ag s uc u e.
In his scena io, a p o ide is likely o be looking o an op imiza ion in he
usage o i s esou ces. Thus, ag eemen offe s should a y in a ce ain pe iod by
he mean ime be ween wo consecu i e eques s (MTBR) wi hin he pe iod.
–The global alidi y o he SLA is om oc obe 1/2007 o decembe 30/2007.
–All Sundays a 23:00h. se e s a e down o an hou due o main enance.
This affec s he global alidi y pe iod.
–The P o ide needs pa o i s se e esou ces o i s own compu ing ne-
cessi ies om Mondays o F idays, om 8:00h o 18:00h. The e o e, in such
ime pe iod, he p o ide equi es ha consume s speci y in hei se ice
eques s a MTBR g ea e o equal o 20 seconds.
–The p o ide allows eques s wi h “Dijks a”, “Knapsack”, o “K uskal” al-
go i hm allowing execu ion es s 48 hou s be o e ag eemen ini ia ion da e
wi h any alue o he MTBR. Bu 24 hou s be o e ag eemen ini ia ion da e
he p o ide cons ains hese es s wi h a minimum MTBR o 40s.
3 Complexi y on De ining Ou DSL
Mos o he ime, design is a challenging ask wi h a high deg ee o unce ain y.
In pa icula , we ace he challenge o c ea ing a DSL o exp ess he empo al
equi emen s ha we could find in eal-wo ld SLAs. A he end o he designing
p ocess we ound ha , choosing he WS-Ag gi es us a consis en amewo k.
In his sec ion, we analyse wi h a mo e c i ical poin o iew, he ad an ages
and d awbacks o ou app oach in o de o sp ead ou know-how in he design
o he empo al ex ension so i could be used as a basis o u he deba e on
empo al-awa e SLAs.
The exp essi eness and succinc ness o a DSL a e key ac o s o conside ing
i as a good p oposal o no . Ou empo al DSL is he esul o se e al wo ks
whe e we ha e al eady s udied empo ali y on web se ices: In [11], we p esen ed
a cons ain -based app oach o empo al-awa e web se ices p ocu emen . In
[12], we elabo a ed a s udy abou exp essi eness o empo al desc ip ions o
web se ices. And in [13], while imp o ing empo al-awa eness o WS-Ag, we
e iewed he kinds o empo al pe iods defined in se e al p oposals om seman ic
and adi ional web se ice specifica ions and addi ionally we s udied he pe iods
defined by he IETF RFC 3060 [20]. We concluded ha he exp essi eness o
ou p oposal is simila o hei s.
In ela ion o succinc ness i is impo an o ema k ha as we show in [13],
we conside he pe iodical alidi y pe iods as a main conce n in he ag eemen s
domain a he han simply using a se ime in e als (choice made by mos o
he app oaches ound in he li e a u e [2,3,4,6,9,10,21,22]). In his con ex , o he
bes o ou knowledge he only p oposal wi h explici pe iodic e ms is IETF RFC
3060 which includes a la ge amoun o p ope ies such as “Mon hO Yea Mask”,

“DayO Mon hMask” and o he syn ac ic s uc u es ha allows he defini ion o
his kind o pe iod as concise as ou defini ion. On he con a y, o he p oposals
ha do no include pe iodic ime pe iods can be defined as less succinc due o
he amoun o effo ha mus be ca ied ou oexp essape iodicbeha iou .As
an example, we would need 1000 pe iod defini ions o desc ibing a single pe iodic
in e al om ou p oposal wi h a alue o 1000.0 o epe i ions p ope y.
The effo o exp ess pe iodic beha iou e en inc eases i he epea ed pe iod
o ime is discon inuous. Ou p oposal allows a simple defini ion o pe iodic and
discon inuous pe iods o ime by mean o a se o in e als defined inside he
disjoin elemen o e which he pe iodici y is defined. Addi ionally, a common
limi a ion ound in o he p oposals is ela ed wi h he possibili y o speci y-
ing es ic ions only o e ce ain pa s o he ag eemen ; his limi a ion can be
desc ibed as a main d awback in he exp essi eness o he DSL.
Ano he impo an issue is ela ed wi h pa ies p e e ences o an SLAs. In
gene al, each o he pa ies ha sign a pa icula ag eemen , ha e o speci y in a
conc e e manne i s p e e ences. In WS-Ag, his is made by means o an ag ee-
men empla e ha should be comple ed o he o he pa y. In his p e e ence
con ex , we ound a li le se o p oposals , among hose which a e empo al-
awa e, ha ha e aken he p e e ences in o accoun . Howe e , o he bes o ou
knowledge, none o hem ha e s udied empo ali y on p e e ences and c ea ion
cons ain s.
We dis inguish wo ways o decla e he p e e ences: (1) by compa ing he “de-
g ee o simila i y” be ween alues o se ice p ope ies om diffe en ag eemen
offe s and empla es; o example, i a p o ide specifies in he ag eemen em-
pla e ha i p e e s a alue o MTBR o 30s, an ag eemen offe which equi es
a MTBR o 32s will be mo e simila o he empla e han ano he offe equi ing
20s; and (2) by compa ing u ili y alues gi en by u ili y unc ions defined on
he se ice p ope ies, jus as we ha e desc ibed abo e. Bo h al e na i es use
weigh s as a means o inco po a ing he deg ee o impo ance among se ice
p ope ies o he p e e ences. UDDIe [2], EWSDL [3] and METEOR-S [17] ha e
based hei p e e ences on he deg ee o simila i y, whe eas o he p oposals ha e
only men ioned hei in e es . Rega ding u ili y unc ions, WSMO/WSML [5],
Gonzalez e al. [6] and METEOR-S [17] a e cu en ly wo king on inco po a -
ing his ea u e o hei p oposals. The e o e in p e e ences ou p oposal is as
exp essi e as mos exp essi e p oposals.
In addi ion, a key poin add essed by ou DSL has been o exp ess a sui able
seman ic o de e mining he consis ency o an ag eemen offe / empla e, he
con o mi y be ween empla es and offe s, and heselec iono heop imaloffe
o a gi en empla e.
In [11], we desc ibed he seman ic o hese ope a ions applied o web se ices
specifica ions and ma chmaking p ocesses bu in his pape we conside hei on
he SLAs es ablishmen con ex . Mo e o mally, we iden i y he ollowing h ee
ope a ions ha a e highly common in aul - ole an scena ios: (i) consis ency (i.e
an ag eemen documen is consis en i i does no ha e con adic o y e ms);
(ii) con o mi y (i.e. wo ag eemen documen s con o m i he gua an ee e ms
om each pa y suppo he equi emen s o he o he pa y); (iii) op imali y,
an ag eemen offe is op imal o a SLA when i s gua an ee e ms gi e he bes
suppo o he equi emen s o he o he pa y and ice e sa.
Finally, while desc ibing ou empo al DSL his has lead us o disco e a wide
ange o uses o empo ali y in WS-Ag, i is impo an o highligh he explici
inclusion o a empo al ea u e made in he specifica ion; ha is he li e ime
o he whole ag eemen . As a complemen , we can summa ise he new ange o
possibili ies when using ou app oach; depending on he a ge we could ha e
empo al-awa e c ea ion cons ain s o p e e ences o e se ice p ope ies:On
he one hand, we can exp ess empo al c ea ion cons ain s in he ag eemen c e-
a ion p ocess, e.g. “P o ide mus allow execu ion es s wi h a minimum MTBR
o 40s, 48 hou s be o e ag eemen ini ia ion da e in o de o c ea e he ag ee-
men ”. In [13], we gi e mo e de ails o empo al c ea ion cons ain s on single o
mul iple cons ain s. On he o he hand, we allow he specifica ion o empo al
p e e ences o e se ice p ope ies.
Addi ionally, we define a schema o using any kind o u ili y unc ions o e
any se o se ice p ope ies, ins ead o only cons an floa unc ions as desc ibed
in WS-Ag. Wi h his app oach, we now o mula e he p e e ences o he pa ies
o e a se ice p ope y (o a se ) e sus o he p ope ies wi h any kind o ma h-
ema ical unc ion, lea ing open he conc e e language o speci y hose unc ions.
4 Discussion
Based on he expe ience o de eloping an imp o emen o empo al-awa eness o
WS-Ag we belie e ha some impo an opics mus be discussed. In his sec ion,
we p opose a se o key poin s ha mus be analysed in o de o deal p ope ly
wi h empo ali y in eal-wo ld scena ios. Main aim o his lis is o es ablish a
ich discussion a he han expose an exhaus i e analysis.
–To de elop conc e e DSL o desc ibing empo al p ope ies in SLAs based
on eal scena ios.
–To ha e a empo al DSL as exp essi e as possible in o de o allow he
desc ip ion o a ich se o empo al cons ain s.
–To exp ess his succinc ly, empo ali y is made easie o i s use allowing i
o be mo e unde s andable and clea e o bo h pa ies.
–To apply he ope a ions o consis ency, con o mi y and op imali y on he
SLA es ablishmen boos s an au oma ion o he p ocess.
–To allow he defini ion o a iable p e e ences based on empo al-awa e u il-
i y unc ions o suppo ing scena ios in which he pa ies p e e ence may
a y a imes, such as “an inc easing u ili y unc ion (x)will be used in
ush hou and a dec easing u ili y unc ion (x)will be applied a any
o he ins an ”. This gi es mo e exp essi eness o mo e succinc ness o ou
p oposal.
–To allow he defini ion o empo al c ea ion cons ain s. Thus, we could
desc ibe cons ain such as “In o de o c ea e he ag eemen you mus ha e
paid he 50% o he o al p ice a week be o e he ag eemen ini a ion da e”.
5 Conclusions
In his pape , we show ou expe ience in he ex ension o WS-Ag as he s a ing
poin o p opose a se o key opics o s a a discussion abou he con enience o
o conside empo al p ope ies as a main conce n in SLAs. As a conclusion we
can enume a e he ollowing issues: We conside he impo ance o ex end he
empo al-awa eness WS-Ag by defining an app op ia e domain-specific language.
Based on he discussion p esen ed, we belie e he empo al DSL mus ha e a
wide exp essi eness in o de o allow as many empo al p ope ies as possible.
Al hough he succinc ness is no essen ial o he DSL i makes i easie and
mo e unde s andable o an use .
Addi ionally, o boos an au oma ion o he p ocess based on p e ious wo ks
[11,19] we should apply he ope a ions o consis ency, con o mi y and op imali y
on he SLA es ablishmen . Fu he mo e, o suppo scena ios in whe e he pa ies
p e e ence which may a y om ime o ime we should allow he defini ion o
a iable p e e ences based on empo al-awa e u ili y unc ions; e.g. he p o ide
would a y i s p e e ences acco ding o he usage le el o he esou ces. These
kinds o p e e ences gi e mo e exp essi eness o ou p oposal in a succinc way
because a complex u ili y unc ion desc ibed on a conc e e pe iod o ime can
be defined as se e al cons an u ili y unc ions on consecu i e ins an s.
Finally, we need o desc ibe empo al cons ain s in he ag eemen c ea ion
p ocess in o de o allow ha a iabili y o he pa ies would a y om ime
o ime oo. This is why a e conce ned abou empo al-awa eness in SLAs. Ou
p oposal is qui e exp essi e, bu wha abou efficiency in he au oma ion o he
p ocess?. So, exp essi eness on one side and efficiency on he o he , can ac ually
ep esen wo sides o he same coin (You can’ ha e one wi hou he o he ). In
o de o ob ain a good solu ion we should ge a good balance be ween hem.
Re e ences
1. Aiello, M., F anko a, G., Mal a i, D.: Wha ’s in an ag eemen ?An analysis and
an ex ension o WS-ag eemen . In: Bena allah, B., Casa i, F., T a e so, P. (eds.)
ICSOC 2005. LNCS, ol. 3826, pp. 424–436. Sp inge , Heidelbe g (2005)
2. Ali, A.S., Al-Ali, R., Rana, O., Walke , D.: UDDIe: An Ex ended Regis y o Web
Se ices. In: P oc. o he IEEE In ’l Wo kshop on Se ice O ien ed Compu ing:
Models, A chi ec u es and Applica ions a SAINT Con e ence. IEEE P ess, Los
Alami os (2003)
3. Chen, Y., Li, Z., Jin, Q., Wang, C.: S udy on qoS d i en web se ices composi ion.
In: Zhou, X., Li, J., Shen, H.T., Ki su egawa, M., Zhang, Y. (eds.) APWeb 2006.
LNCS, ol. 3841, pp. 702–707. Sp inge , Heidelbe g (2006)
4. Chen, Z., Liang-Tien, C., Bu-Sung, L.: Seman ics in Se ice Disco e y and QoS
Measu emen . IT P o - IEEE Compu e Socie y, 29–34 (2005)
5. de B uijn, J., Feie , C., Kelle , U., La a, R., Polle es, A., P edoiu, L.: WSML
Reasoning Su ey (No embe 2005)
6. González-Cas illo, J., T as ou , D., Ba olini, C.: Desc ip ion Logics o Ma ch-
making o Se ices. Technical Repo HPL-2001-265, Hewle -Packa d (2001)
7. OGF G id Resou ce Alloca ion Ag eemen P o ocol WG (GRAAP-WG). Web Se -
ices Ag eemen Speci ica ion (WS-Ag eemen ), . g d.107 (2007)
8. IBM. Web Se ice Le el Ag eemen (WSLA) Language Speci ica ion (2003)
9. Li, L., Ho ocks, I.: A So wa e F amewo k o Ma chmaking based on Seman ic
Web Technology. In: P oc. o he 12 h ACM In l. Con . on WWW, pp. 331–339
(2003)
10. Lodi, G., Panzie i, F., Rossi, D., Tu ini, E.: SLA-D i en Clus e ing o QoS-Awa e
Applica ion Se e s. IEEE T ansac ions on So wa e Enginee ing 33(3), 186–196
(2007)
11. Ma ín-Díaz, O., Ruiz-Co és, A., Du án, A., Mülle , C.: An app oach o empo al-
awa e p ocu emen o web se ices. In: Bena allah, B., Casa i, F., T a e so, P.
(eds.) ICSOC 2005. LNCS, ol. 3826, pp. 170–184. Sp inge , Heidelbe g (2005)
12. Mülle , C., Ma ín-Díaz, O., Resinas, M., Fe nández, P., Ruiz-Co és, A.: A WS-
Ag eemen Ex ension o Speci ying Tempo al P ope ies in SLAs. In: P oc. o he
3 d Jo nadas Cien í ico-Técnicas en Se icios Web y SOA (2007)
13. Mlle , C., Ma ín-Díaz, O., Ruiz-Co és, A., Resinas, M., Fe nández, P.: Imp o ing
empo al-awa eness o WS-ag eemen . In: K äme , B.J., Lin, K.-J., Na asimhan,
P. (eds.) ICSOC 2007. LNCS, ol. 4749, pp. 193–206. Sp inge , Heidelbe g (2007)
14. Di Modica, G., Toma chio, O., Vi a, L.: A amewo k o he managemen o dy-
namic SLAs in composi e se ice scena ios. In: Di Ni o, E., Ripeanu, M. (eds.)
ICSOC 2007. LNCS, ol. 4907, pp. 139–150. Sp inge , Heidelbe g (2007)
15. Molina-Jiménez, C., P uyne, J., an Moo sel, A.: The ole o ag eemen s in IT
managemen so wa e. In: de Lemos, R., Gacek, C., Romano sky, A. (eds.) A chi-
ec ing Dependable Sys ems III. LNCS, ol. 3549, pp. 36–58. Sp inge , Heidelbe g
(2005)
16. OASIS and UN/CEFAT. Elec onic business using XML, ebXML (2007)
17. Oldham, N., Ve ma, K., She h, A., Hakimpou , F.: Seman ic WS-Ag eemen Pa -
ne Selec ion. In: 15 h In e na ional WWW Con ., pp. 697–706. ACM P ess, New
Yo k (2006)
18. Papazoglou, M.P., Geo gakopoulos, D.: Se ice o ien ed compu ing. Commun.
ACM 46(10), 24–28 (2003)
19. Ruiz-Co és, A., Ma ín-Díaz, O., Du án, A., To o, M.: Imp o ing he Au oma ic
P ocu emen o Web Se ices using Cons ain P og amming. In . Jou nal on Co-
ope a i e In o ma ion Sys ems 14(4) (2005)
20. The In e ne Socie y. Policy Co e In o ma ion Model - 1 Speci ica ion (2001)
21. Tian, M., G amm, A., Naumowicz, T., Ri e , H., Schille , J.: A Concep o QoS
In eg a ion in Web Se ices. In: P oc. o he IEEE In ’l Web Se ices Quali y
Wo kshop (WISE 2003), pp. 149–155 (2003)
22. T as ou , D., Ba olini, C., González-Cas illo, J.: A Seman ic Web App oach o
Se ice Desc ip ion o Ma chmaking o Se ices. Technical Repo HPL-2001-183