An Initial Approach to Explaining SLA Inconsistencies

An Ini ial App oach o Explaining SLA Inconsis encies? Ca los Mülle , An onio Ruiz-Co és, Manuel Resinas 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, esinas}@us.es Abs ac An SLA signed by all in e es ed pa ies mus be c ea ed ca e- ully, a oiding con adic ions be ween e ms, because hei e ms could ca y penal ies in case o ailu e. Howe e , his consis ency checking may become a challenging ask depending on he complexi y o he ag ee- men . As a consequence, an au oma ed way o checking he consis ency o an SLA documen and e u ning he se o inconsis en e ms o he ag eemen would be e y appealing om a p ac ical poin o iew. Fo ins ance, i enables he de elopmen o so wa e ools ha make he c e- a ion o co ec SLAs and he consis ency checking o impo ed SLAs easie o use s. In his pape , we p esen he p oblem o explaining WS- Ag eemen inconsis encies as a cons ain sa is ac ion p oblem (CSP), and hen we use a CSP sol e oge he wi h an explana ion engine o check he consis ency and e u n he inconsis en e ms. Fu he mo e, a p oo -o -concep using Choco sol e in conjunc ion wi h he Palm expla- na ion engine has been de eloped. Keywo ds: Se ice Le el Ag eemen , WS-Ag eemen , Consis ency Check- ing, Debugging, Quali y o Se ice. 1 In oduc ion SLAs consis o a se o e ms ha include in o ma ion abou unc ional ea u es, non- unc ional gua an ees, compensa ion, e mina ion e ms and any o he e ms wi h ele an in o ma ion o he ag eemen . An ag eemen signed by all in e - es ed pa ies should be edac ed ca e ully because a ailu e o speci y hei e ms could ca y penal ies o he ini ia ing o esponding pa y. The e o e, ag eemen e ms should be speci ied in a consis en way, a oiding con adic ions be ween hem. Howe e , depending on he complexi y o he ag eemen , his may be- come a challenging ask. Fo ins ance, in a scena io in which a p o ide o e s compu ing se ices o o he o ganiza ions, an SLA could be ag eed ha includes non- unc ional a ibu es such as: he a ailabili y -in pe cen age-, he mean ime be ween wo consecu i e eques s o he se ice (MTBR) -in seconds-, and he ?This wo k has been pa ially suppo ed by he Eu opean Commission (FEDER), Spanish Go e nmen unde CICYT p ojec Web-Fac o ies (TIN2006-00472), and p ojec P07-TIC-2533 unded by he Andalusian local Go e nmen . e iciency. Assuming ha : A ailabili y anges be ween [90..100] , MTBR anges be ween ∈[5..60] and E iciency =A ailabili y/MTBR. I he SLA includes a e m obliga ing o gua an ee E iciency >20, a i s sigh he SLA is consis en . Howe e he highes alid alue o e iciency is 100/5=20, so his e m canno be sa is ied. The e o e, a consis ency checke ha au oma ically checks he SLA o inconsis encies be ween i s e ms would be e y appealing om a p ac ical poin o iew. Fu he mo e, i is o in e es no only o ob ain an au oma ed way o check- ing he consis ency o an SLA documen , bu also o e u n an explana ion i he documen is inconsis en . This explana ion is he se o inconsis en e ms o he ag eemen . So, in he p e ious scena io we would ob ain as debugging in o ma- ion ha he inconsis en e ms a e: [(E iciency >20),(A ailabili y ∈[90..100]), (MTBR ∈[5..60]),(E iciency =A ailabili y/MTBR)], because hey a e incon- sis en e ms. This au oma ed consis ency checking enables he implemen a ion o a so wa e ool which makes he c ea ion o co ec SLAs and he consis ency checking o impo ed SLAs easie o use s. Ne e heless, as a as we know, he consis ency o SLAs has been aken o g an ed by mos au ho s. In his pape , we desc ibe a mechanism o check and explain SLAs speci ied wi h WS-Ag eemen [2], which is a p oposed ecommen- da ion o he Open G id Fo um (OGF) which p o ides a schema o de ining SLAs and a p o ocol o c ea ing hem. To his end, we map he e ms o a subse o he WS-Ag eemen documen in o a cons ain sa is ac ion p oblem (CSP). Then we use he CSP as an inpu o a CSP sol e wi h an explana ion engine, which will e u n he se o inconsis en cons ain s. Finally, we ace back he cons ain s o he o iginal SLA e ms in o de o gi e use ul debugging in o ma ion o use s. As a p oo -o -concep , we ha e de eloped a p o o ype o ou consis ency analyse using Choco sol e [1] in conjunc ion wi h he Palm explana ion engine [3]. This p o o ype is a ailable o es ing a h p://www.isa.us.es in he ools sec ion. This pape is s uc u ed as ollows. Sec ion 2 p esen s some backg ound on cons ain sa is ac ion p oblems and WS-Ag eemen . Sec ion 3 de ails he sub- se o WS-Ag eemen which is used o explain he SLA inconsis encies in 3.1 and i s mapping o CSP in 3.2. Sec ion 4 desc ibes ou p ocess o explain he WS-Ag eemen * iconsis encies. Sec ion 5 shows ou p oo -o -concep . Sec ion 6 epo s on he ela ed p oposals. Finally, Sec ion 7 de ails ou conclusions and u u e wo k. 2 P elimina ies 2.1 Cons ain Sa is ac ion P oblems Cons ain Sa is ac ion P oblems [8] ha e been he objec o esea ch in A i icial In elligence o e he las ew decades. A Cons ain Sa is ac ion P oblem (CSP) is de ined as a se o a iables, each anging on a ini e domain, and a se o cons ain s es ic ing all he alues ha hese a iables can ake simul aneously. A solu ion o a CSP is an assignmen o a alue om i s domain o e e y a iable, in such a way ha all cons ain s a e sa is ied simul aneously. These a e some basic de ini ions o wha a CSP is. De ini ion 1 (CSP). A CSP is a h ee– uple o he o m (V, D, C)whe e V6=∅ is a ini e se o a iables, D6=∅is a ini e se o domains (one o each a iable) and Cis a cons ain de ined on V. Conside , o ins ance, he CSP: ({a, b},{[0..2],[0..2]},{a+b < 4}) De ini ion 2 (Solu ion). Le ψbe a CSP, a solu ion o ψis wha e e alid assignmen o all elemen s in V ha sa is ies C. In he p e ious example, a possible solu ion is (2,0) since i e i ies ha 2+0<4. De ini ion 3 (Solu ion space). Le ψbe a CSP o he o m (V, D, C), i s solu ion space deno ed as sol(ψ)is made up o all i s possible solu ions. A CSP is sa is iable i i s solu ion space is no emp y. sol(ψ) = {S| ∀si·si∈S⇒C(si) = ue} In he p e ious example he e a e eigh solu ions. The only assignmen ha does no sa is y a+b < 4is (2,2). Ne e heless, i we eplace he cons ain wi h a+b < −1, hen he CSP is no sa is iable. In many eal-li e applica ions, i a CSP has no solu ion, we would like o know which se o cons ain s a e esponsible o his si ua ion. This can be done by in e p e ing he CSP as an explana ion p oblem. De ini ion 4 (Explana ion p oblem). Le be a CSP o he o m (V, D, C) wi h an emp y solu ion space: sol() = ∅. I is conside ed o be an explana ion p oblem i i s objec i e is o ind a se o cons ain s C0⊂C ha canno be sa is ied. De ini ion 5 (Explana ions). Le be an explana ion p oblem, he esul ing se o inconsis en cons ain s C0a e known as he explana ions o he p oblem. They a e di ided in o wo pa s: a subse o he o iginal se o cons ain s C0⊂C and a subse o decision cons ain s in oduced so a in he sea ch o solu ions (dc1, ..., dck). As de ined in [3], a con adic ion explana ion, also known as “nogood” [7], is a subse o he cons ain s o he p oblem ha , le alone, leads o a con adic ion (no easible solu ion con ains a nogood). The p e ious CSP example, wi h he cons ain eplaced wi h a+b < −1, is no sa is iable, and by in e p e ing i as an explana ion p oblem he explana ion engine should ob ain as he se o inconsis en cons ain s: [(a+b < −1),(a >= 0),(b >= 0)] 2.2 WS-Ag eemen in a Nu shell WS-Ag eemen speci ies an XML-based language and a p o ocol o ad e ising he capabili ies and p e e ences o se ice p o ide s, and c ea ing ag eemen s based on ag eemen o e s. The s uc u e o an ag eemen in WS-Ag eemen comp ises: – Name: i iden i ies 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 as ini ia o o esponde in he ag eemen . Addi ionally, i can include o he impo an in o ma ion o he ag eemen . – Te ms: hey a e g ouped by he ollowing e m composi o s: Exac lyOne, OneO Mo e, o All. The wo main ypes o e ms a e: •Se ice e ms: hey p o ide se ice in o ma ion by means o : ∗Se ice desc ip ion e ms and se ice e e ences, which includes in o - ma ion o ins an ia e o iden i y he se ices and ope a ions in ol ed in he ag eemen . ∗Se ice p ope ies, which includes he measu able p ope ies ha a e used in exp essing gua an ee e ms. They consis o a se o a i- ables whose alues can be es ablished inside he se ice desc ip ion e m, and whose domain can be es ablished by he me ic a ibu e poin ing o an ex e nal XML documen . •Gua an ee e ms: hey desc ibe he se ice le el objec i es (SLO) ag eed by a speci ic obliga ed pa y, ei he using a ee- o m elemen o using a key pe o mance indica o . I also includes he scope o he e m (e.g. i i applies o a ce ain ope a ion o a se ice o he whole se ice i sel ), and a quali ying condi ion ha speci ies he alidi y condi ion unde which he e m is applied. Figu e 1 depic s an example o a WS-Ag eemen be ween a compu ing se - ices p o ide and a consume . I de ines se e al se ice p ope ies whose domain is speci ied in an ex e nal XML documen (depic ed in Figu e 2). No e ha o he XML documen s, such as an XML Schema de ini ion, could ha e been used in- s ead. The se ice p ope ies de ined in he WS-Ag eemen documen o Figu e 1 a e he ollowing ones: – he a ailabili y -in ege o m 1 o 100- – he mean ime be ween wo consecu i e eques s o he se ice (MTBR) - in ege g ea e han 1- – he mean ime o esponse (MTTR) -in ege g ea e han 1- – he ini ial cos o he se ice (Ini Cos ) -in ege g ea e han 1- – he inal cos o he se ice (Cos ) -in ege g ea e han 1- – he inc ease o he cos i he MTBR < 10 (Ex aMTBRCos ) -in ege g ea e han 1- – he inc ease o he cos i he MTTR < 05 (Ex aMTTRCos ) -in ege g ea e han 1- We ha e ex ac ed he ollowing in o ma ion om he SLA: –MTBR ∈[5..60]. –MTTR ∈[1..10]. –I MTBR >= 10 Then A ailabili y ∈[90..100]. –I MTBR < 10 Then A ailabili y ∈[95..100]. –Cos = Ini Cos + Ex aMTBRCos + Ex aMTTRCos . –I MTBR < 10 Then Ex aMTBRCos = 15. –I MTTR < 05 Then Ex aMTTRCos = 15. –I MTBR >= 10 and MTTR >= 05 Then Ex aMTBRCos = 0 and Ex- aMTTRCos = 0. 3 Mapping WS-Ag eemen on o CSP 3.1 WS-Ag eemen * as a subse o WS-Ag eemen Due o he lexibili y and ex ensibili y o WS-Ag eemen , we ocus in his pape on a subse o WS-Ag eemen ha is a bi less exp essi e han WS-Ag eemen , bu s ill use ul o ou pu pose. The subse o WS-Ag eemen is called WS- Ag eemen *. A WS-Ag eemen * documen (α) is composed o he ollowing el- emen s: –Se ice p ope ies mus de ine all a iables ha a e used in he gua an ee e ms. In addi ion, a ibu e me ic o he a iables is manda o y. This a ibu e mus poin o ano he XML documen o schema ha p o ides he da a ype and a gene al ange o alues (δ) o each se ice p ope y. Figu e 2 shows an example o an ad-hoc XML documen ha includes he men ioned in o ma ion o se ice p ope ies o he example o Figu e 1, al hough o he XML documen s could be used. XML node Loca ion (λυ) o each a iable es ablishes he speci ic XML node inside he se ice desc ip ion e m whe e i is de ined he alue o such a iable ( alue(λυ)). –Te ms (T) can be composed using he h ee e m composi o s de ined in WS-Ag eemen : All (), Exac lyOne (⊗), and OneO Mo e (⊕). –Se ice desc ip ion e ms can be included bu only o impose speci ic alue de ini ions o each a iable (υ) o se ice p ope ies. O he se ice desc ip- ion e ms could be added bu hey a e igno ed when checking he SLA consis ency. –Gua an ee e ms (γ) can be included. Bo h quali ying condi ion (κ) and he SLOs (σ) mus be de ined as cons ain s on he a iables de ined in he se ice p ope ies, and only o hose a iables (i.e. κ∈Cand σ∈C). The scope o a gua an ee e m canno be es ablished, bu all gua an ee e ms apply o he whole se ice. No e ha , al hough WS-Ag eemen * is no as exp essi e as WS-Ag eemen , i does allow o he exp ession o complex ag eemen s. Fo ins ance, he ag ee- men depic ed in Figu e 1 is compa ible wi h WS-Ag eemen *. Thus, we can o mally de ine an ag eemen speci ied wi h WS-Ag eemen * as ollows: <Ag eemen ’’id=exampleScena io’’> <Con ex > ... </...> <All> <Se iceDesc ip ionTe m Name=’’Compu ingSe ice’’> <...> ... </...> <Ini Cos > 20 </...> ... </...> </...> </Se iceDesc ip ionTe m> <Se iceP ope ies> ...<Va iable name=’’A ailabili y’’ me ic=’’me icXML:A ailabili y’’> <Loca ion> A ailabili y </Loca ion> </Va iable> <Va iable name=’’MTBR’’ me ic=’’me icXML:MTBR’’> <Loca ion> MTBR </Loca ion> </Va iable> <Va iable name=’’MTTR’’ me ic=’’me icXML:MTTR’’> <Loca ion> MTTR </Loca ion> </Va iable> <Va iable name=’’Ini Cos ’’ me ic=’’me icXML:Ini Cos ’’> <Loca ion> Ini Cos </Loca ion> </Va iable> <Va iable name=’’Cos ’’ me ic=’’me icXML:Cos ’’> <Loca ion> Cos </Loca ion> </Va iable> <Va iable name=’’Ex aMTBRCos ’’ me ic=’’me icXML:Ex aMTBRCos ’’> <Loca ion> Ex aMTBRCos </Loca ion> </Va iable> <Va iable name=’’Ex aMTTRCos ’’ me ic=’’me icXML:Ex aMTTRCos ’’> <Loca ion> Ex aMTTRCos </Loca ion> </Va iable>... </Se iceP ope ies> <Gua an eeTe m Name=’’MTBRDomain’’> <SLO> MTBR >= 5 and MTBR <= 60 </...> </Gua an eeTe m> <Gua an eeTe m Name=’’MTTRDomain’’> <SLO> MTTR >= 1 and MTTR <= 10 </...> </Gua an eeTe m> <Gua an eeTe m Name=’’Cos De ini ion’’> <SLO> Cos = Ini Cos + Ex aMTBRCos + Ex aMTTRCos </...> </Gua an eeTe m> <Exac lyOne> <Gua an eeTe m Name=’’Lowe A ailabili yDomain’’> <Quali yingCondi ion> MTBR >= 10 </...> <SLO> A ailabili y >= 90 and A ailabili y <= 100 </...> </Gua an eeTe m> <Gua an eeTe m Name=’’Highe A ailabili yDomain’’> <Quali yingCondi ion> MTBR < 10 </...> <SLO> A ailabili y >= 95 and A ailabili y <= 100 </...> </Gua an eeTe m> </Exac lyOne> <OneO Mo e> <Gua an eeTe m Name=’’MTBRInc emen ’’> <Quali yingCondi ion> MTBR < 10 </...> <SLO> Ex aMTBRCos = 15 </...> </Gua an eeTe m> <Gua an eeTe m Name=’’MTTRInc emen ’’> <Quali yingCondi ion> MTTR < 05 </...> <SLO> Ex aMTTRCos = 15 </...> </Gua an eeTe m> <Gua an eeTe m Name=’’Cheape Cos ’’> <Quali yingCondi ion> MTBR >= 10 and MTTR >= 05 </...> <SLO> Ex aMTBRCos = 0 and Ex aMTTRCos = 0 </...> </Gua an eeTe m> </OneO Mo e> </All> </Ag eemen > Figu e 1. Example o a WS-Ag eemen documen wi h all kinds o composi o s. <me icXML> <A ailabili y ype=’’in ege ’’ min=’’1’’ max=’’100’’/> <MTBR ype=’’in ege ’’ min=’’1’’ max=’’unbounded’’/> <MTTR ype=’’in ege ’’ min=’’1’’ max=’’unbounded’’/> <Ini Cos ype=’’in ege ’’ min=’’1’’ max=’’unbounded’’/> <Cos ype=’’in ege ’’ min=’’1’’ max=’’unbounded’’/> <Ex aMTBRCos ype=’’in ege ’’ min=’’1’’ max=’’unbounded’’/> <Ex aMTTRCos ype=’’in ege ’’ min=’’1’ max=’’unbounded’’/> </me icXML> Figu e 2. Ad-hoc XML documen o he a iable domains o Figu e 1. De ini ion 6 (A WS-Ag eemen * documen ). A WS-Ag eemen * docu- men αis a se o a iables υi, a iable domains δi, and a se o e ms T, including se ice desc ip ion e ms, gua an ee e ms and e ms composi o s as ollows: α= (υi, . . . , υn, δi, . . . , δn, T1, . . . , Tm),whe e Ti=            λυ γ Ti1. . . Tik Ti1⊗. . . ⊗Tik Ti1⊕. . . ⊕Tik            3.2 Mapping WS-Ag eemen * on o CSP Figu e 3 depic s he mapping (µ) o a WS-Ag eemen * documen (α) on o an equi alen CSP, (ψα). The a iables (υ) de ined inside he se ice p ope ies a e he CSP a iables; he a iable domains (δ) included in he documen speci ied by he me ic a ibu e a e he CSP a iable domains; and he cons ain s om he se ice desc ip ion e ms (λυ), gua an ee e ms (γ) and e m composi o s (as a logic “AND”, ⊗as logic “XOR”, and ⊕as logic “OR”) a e he CSP cons ain s. Thus, in gene al, ou WS-Ag eemen * o CSP mapping can be de ined as ollows: De ini ion 7 (Mapping an WS-Ag eemen * o CSP). The mapping (µ: α→ψ) o a WS-Ag eemen * documen (α) o a CSP (ψ) can be de ined as ollows: µ(α) = µ(υ1, . . . , υn, δ1, . . . , δn, T1, . . . , Tm) = = ({υ1, . . . , υn},{δ1, . . . , δn},{µT(T1, . . . , Tm)}) = ψα whe e µT:T→Cis a mapping unc ion o e ms in o cons ain s de ined as ollows: µT(T)≡            υ=λυi Tis a se ice desc ip ion e m (λυ) σi Tis a gua an ee e m wi hou quali ying condi ion (γσ) κ⇒σi Tis a gua an ee e m wi h a quali ying condi ion (γσ,κ) Vk i=1 µT(Ti)i Tis a composi e e m (T1. . . Tk) Vk i=1 µT(Ti)⇔(Vk j=1 j6=i¬µT(Tj)) i Tis a composi e e m (T1⊗. . . ⊗Tk) Wk i=1 µT(Ti)i Tis a composi e e m (T1⊕. . . ⊕Tk)            <Ag eemen > <o o > <Se iceP ope ies> …< a iable name=‘’ 1’’ me ic=‘‘me icXML:δ1’’> <loca ion> 1 </…> </ a iable> … …< a iable name=‘‘ n’’ me ic=‘‘me icXML:δn’’> <loca ion> n </…> </ a iable> … </Se iceP ope ies> <Se iceDesc ip ionTe m ( 1)> …< 1> alue( 1)</ 1> … <n> alue( n)</ n> … </Se iceDesc ip ionTe m> <Gua an eeTe m name=‘‘ 1’’ ( 2)> <SLO> </SLO> </Gua an eeTe m> <Gua an eeTe m name=‘‘ m’’ ( k)> <Quali yingCondi ion> </Quali yingCondi ion> <SLO> </SLO> </Gua an eeTe m> </o o > </Ag eemen > <<WS-Ag eemen *>> <me icXML> <δ1 ype=‘‘…’’ min=‘‘…’’ max=‘‘…’’> … <δn ype=‘‘…’’ min=‘‘…’’ max=‘‘…’’> </me icXML> <<me icXML>> CSP a iables ={ (Equi alen CSP) CSP a iable domains CSP cons ain s 1, …, n, δ1, …, δn, 1, …, m (i)≡ i = alue( i) (1)≡ i≡ i (m)≡ (o o ) δi≡ δi { { { { { ( ( Figu e 3. Summa y o WS-Ag eemen * o CSP mapping. Using his mapping, he ψα o he example o Figu e 1 is mapped as ollows: ψα= ( { A ailabili y, MTBR, MTTR, Ini Cos , Cos , Ex aMTBRCos , Ex aMTTRCos }, {[1. . . 100 ], [ 1 . . . ∞), [ 1 . . . ∞), [ 1 . . . ∞), [ 1 . . . ∞), [ 1 . . . ∞), [ 1 . . . ∞), }, { Ini Cos = 20, MTBR ≥5 and MTBR ≤60, MTTR ≥1 and MTBR ≤10, Cos = Ini Cos + Ex aMTBRCos + Ex aMTTRCos , ((MTBR ≥10) ⇒(A ailabili y ≥90 and A ailabili y ≤100)) ⇔ ⇔ ¬ ((MTBR < 10) ⇒(A ailabili y ≥95 and A ailabili y ≤100)), (MTBR < 10 ⇒Ex aMTBRCos = 15) ∨ ∨(MTTR < 5 ⇒Ex aMTTRCos = 15) ∨ ∨((MTBR ≥10 and MTTR ≥5) ⇒Ex aMTBRCos = 0 and Ex aMTTRCos = 0) } ) The ollowing able deno es cons ain s mapped om he example o Figu e 1. Example e m (Ti) name Equi alen mapped (µT(Ti)) Ini Cos Ini Cos = 20 MTBRDomain MTBR >= 5 and MTBR <= 60 MTTRDomain MTTR >= 1 and MTTR <= 10 Cos De ini ion Cos = Ini Cos + Ex aMTBRCos + Ex aMTTRCos Lowe A ailabili yDomain (MTBR ≥10) ⇒(A ailabili y ≥90 and A ailabili y ≤100) Highe A ailabili yDomain (MTBR < 10) ⇒(A ailabili y ≥95 and A ailabili y ≤100) MTBRInc emen (MTBR < 10) ⇒(Ex aMTBRCos = 15) MTTRInc emen (MTTR < 5) ⇒(Ex aMTTRCos = 15) Cheape Cos (MTBR ≥10 and MTTR ≥5) ⇒ (Ex aMTBRCos = Ex aMTTRCos = 0) Exac lyOne (Lowe A ailabili yDomain ⇔ ¬ Highe A ailabili yDomain) ∧ (Highe A ailabili yDomain ⇔ ¬ Lowe A ailabili yDomain) OneO Mo e MTBRInc emen ∨MTTRInc emen ∨Cheape Cos All Ini Cos ∧MTBRDomain ∧MTTRDomain ∧ ∧Cos De ini ion ∧Exac lyOne ∧OneO Mo e Table 1. Mapping example e ms o CSP cons ain s 4 Checking and Explaining WS-Ag eemen * Inconsis encies Checking he consis ency o WS-Ag eemen documen s in ol es bo h syn ac ic and seman ic checking. The o me in ol es checking he documen agains he WS-Ag eemen XML Schema. The la e , howe e , is ha de o check on and is hus he ocus o his pape . Figu e 4 depic s ou consis ency checking p ocess. In his scena io, i is im- posed by a human e o , an MTBR alue de ini ion inside he se ice desc ip ion e m wi h a alue o 61. Thus, he ψαwould no be sa is iable by he MTBR domain de ini ion. As depic ed in Figu e 4, we p opose o use he ag eemen speci ied wi h WS-Ag eemen * in conjunc ion wi h he XML documen which de ines he a iables me ics as he wo inpu s o a mapping componen which implemen s he mapping de ined in sec ion 3.2. Once he ag eemen is mapped o a CSP, he explana ion engine o he CSP explaine componen will ob ain whe he he SLA documen is consis en o no . In he la e case, he componen sends o a acing componen he explana ions o he p oblem. In his case, he explana ions o he p oblem a e MT BR: [MTBR = 61; MTBR ≥5 and MTBR ≤60]. The acing componen con e s he explana ions in o he equi alen in- consis en o iginal e ms in o de o gi e use ul in o ma ion o use s. This is done by naming cons ain s mapped om an SLO as he name o he gua an ee e m which includes i ; and i he cons ain was mapped om a alue de ini ion inside a se ice desc ip ion e m, we name i “SDT”, conca ena e wi h he name o he a iable. Then, he p e ious explana ion o MTBR would be aced by us o he inconsis en e m “SDTMTBR” cons ain , showing ha i is he MTBR