logic based mo deling and analysis of w ork o ws a cm
play

' $ Logic Based Mo deling and Analysis of W orko ws A CM - PowerPoint PPT Presentation

' $ Logic Based Mo deling and Analysis of W orko ws A CM PODS'98 | | Hasan Da vulcu* Dept of CS Univ ersit y at Ston y Bro ok N.Y. 11794, U.S.A. * Join t with M. Kifer, C.R. Ramakrishnan, I.V.


  1. ' $ Logic Based Mo deling and Analysis of W ork�o ws A CM PODS'98 | | Hasan Da vulcu* Dept of CS Univ ersit y at Ston y Bro ok N.Y. 11794, U.S.A. * Join t with M. Kifer, C.R. Ramakrishnan, I.V. Ramakrishnan & % 1 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  2. ' $ W ork�o ws W ork�o w: A collection of in ter-related tasks and transactions designed to carry out a w ell-de�ned business pro cess. W ork�o w Managemen t: Automated of w ork, c o or dination among pro cessing en tities, to ac hiev e al . an over al l business go W ork�o w Managemen t System (WfMS): System for automation of w ork�o w pro cesses (lik e DBMS facilitates creation and main tenance of large data sets). & % 2 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  3. ' $ What Kind of Computations are W ork�o ws ? � Long-running: Hours, da ys, mon ths; � Autonomous, distributed pro cessing en tities ; � T ransactional Seman tics. & % 3 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  4. ' $ Bid Ev aluation Example Control Flow Graph: ❏ Contractor Financial Evaluation Analysis o f O.Eval = {Low, High} Receive D.Final = Accept Bid AND C.Cost < Budget XOR s d r c b Store Final Cost Budget Bid D.Final = Reject Decision Update Update Risk Technical Evaluation Analysis t i OR e m External Consultant Consultant Billing Global Coordination Dependencies: ❏ 1. IF o.eval = low THEN not e 2. IF occurs (e) THEN o before e 3. IF occurs (t) AND occurs (e) THEN e before i & % 4. c before f 4 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  5. ' $ Cen tral W ork�o w Problems b 1. d before c XOR a d 2. b AND IF occurs (b) THEN c c Consistency: Is con trol �o w graph and co ordination dep endencies c onsistent ? Correctness: Do es the sp eci�cation satisfy certain k ey prop erties ? e.g. , Ev ery bid is either accepted or rejected b y the bid-ev aluation w ork�o w. Co ordination: Ho w to sc hedule tasks automatically ? & % 5 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  6. ' $ What is Needed to Enable ? Work�ow A utomation Creation: A to sp ecify the structure of pro cesses formal language and their in teractions at a high-lev el; V eri�cation: to ascertain that a w ork�o w is R e asoning metho ds correct; Execution: T ec hniques for correct automatic al ly deriving executions from high-lev el sp eci�cations. � Ideally , all should �t in a single, unifying framew ork. & % 6 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  7. ' $ Requiremen ts for W ork�o w Sp eci�cation � Primitiv es for mo dular sp eci�cation { Se quential and c oncurr ent comp osition; { (e.g. sub-w ork�o ws); Subr outines { De clar ative queries (e.g. transition conditions); { tomicity ; A { Isolation ; � for expressing co ordination dep endencies. T emp or al c onstr aints � (e.g. exceptions) T riggers & % 7 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  8. ' $ Related Researc h on W ork�o ws Extended T ransaction Mo dels � { P ermits hierarc hical structures and relaxes A CID prop erties. [Elmagarmid-90] T ask Dep endencies � { Declarativ e constructs suitable for sp ecifying global co ordination dep endencies. [Klein-91, Singh-96] Activ e Rules and T riggers � { Suitable for sp ecifying reactiv e b eha vior. [Da y al et.al.-91,96] P etri Nets and T emp oral Logics � { Suitable for mo deling and v erifying c oncurr ent pr o c esses . [Sheth et.al.-93, Nabil et.al.-98] & % 8 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  9. ' $ Our F ramew ork: Concurren t T ransaction Logic ( C T ) R : Conserv ativ e extension of �rst order logic for programming , � C T R executing and reasoning with state-c hanging concurren t pro cesses. � C T R uniformly mo dels: { Declarativ e queries { Concurren t up dates { T ransactional b eha vior { Mo del The ory: � A mo del assigns transaction form ulas truth values o v er p aths . (i.e., �nite sequences of states) � Informally , � b eing true o v er path � means \ � can execute along � ." { Pr o of The ory: � Sound and complete. � Constructs exe cution p aths as it pro v es statemen ts. & % 9 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  10. ' $ C T : Logical Connectiv es R � cost up date � budget up date means, \Execute cost update and then execute update ". budg et D D D D D D D D 6 1 2 3 4 5 7 8 cost_update budget_update � ( con tractor � �nancial ) j ( consultan t � billing ) means, \Execute ( contr � ) and ( consul � ) in an interle ave d actor f inancial tant bil l ing fashion". D D D D D D D D 2 3 4 5 6 7 8 1 contractor consultant financial billing ( contractor financial ) ( consultant billing ) & % 10 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  11. ' $ C T : Logical Connectiv es R � bid ev al ^ c b efore f means, \Execute bid ev al and c bef or e f along the same path". (i.e. ^ can b e used to express constrain ts on execution.) c_before_f D D D D D D D D 6 1 2 3 4 5 7 8 bid_eval � _ means, \Execute either in ternal ev al external con tractor inter nal ev al or ". exter nal contr actor � : means, \Do an ything but an execution of external con tractor ". exter nal contr actor � � means, \An execution of db up dates cost up date budget up date cost � is also an execution of updates ". update budg et update db ( is de�ned as _ : � ). � � � & % � � db up dates means, \Execute db updates in isolation". 11 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  12. ' $ Putting It Altogether : Bid-Ev aluation W ork�o w � translates straigh tforw ardly in to CTR form ulas: Con trol-�o w graphs � ( f j j ) � bid ev al r inancial db updates technical r est � ([ o:ev = \ hig h "] � ) _ ( l � ) f inancial o al f ow f � ( c � [ c:cost et ] � b ) � db updates < budg s technical ( t � i ) _ ( e � m ) _ ( t � i j e � m ) � � �� � Global Co ordination Dep endencies can b e mo deled as conjunction of a set of dep endencies C . 1 : O l ow ! : O e 3 : O t ^ O e ! O e � O i 2 : O e ! ( O o � O e ) 4 : O c � O f ( means \ � o c curs sometime".) O � � W ork�o w sp eci�cation: ^ C bid ev al & % 12 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  13. ' $ T emp oral Dep endencies 1. ts : O e , : O e ( e m ust/m ust not o ccur); Existence constrain 2. ts : � ( e m ust o ccur b efore ); Serial constrain O e O f f 3. ts : ^ , and _ , if , are constrain ts.. Complex constrain C C C C C C 1 2 1 2 1 2 R � Constrain ts are C T form ulas. & % 13 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  14. ' $ T ransformation Apply � Executing a w ork�o w G ^ C b y the general pro of theory has run-time complexit y . exp onential � W e dev elop a re-write system, , whic h transforms G ^ C Apply in to an equiv alen t ^ -free form ula G suc h that: C { and b ecomes more e�cien t on G . Exe cution veri�c ation C & % 14 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  15. ' $ W ork�o w Analysis with C T R Prop osition ( Apply ): 8 G if satis�able and G is ^ -free < C C ( C ) C ^ G � ; G � Apply if is unsatis�able f al se C ^ G : A legal execution of can b e pic k ed from in linear time b y the � C ^ G G C inference system. 2 { Inconsistency: I� ( C , ) bid ev al � f al se Apply { V eri�cation: Giv en a prop ert y �: 8 if � holds for G f al se < C ( : � ; ) G � Apply C all coun ter executions of � G C ^: � : and reduce to the same logical problem { � Inc onsistency veri�c ation & % unsatis�abilit y . 15 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

  16. ' $ T ransformation: Con td. Apply α XOR γ β δ η Example: If T is � � ( � _ � _ � ) � � , then ( O � ) = ^ = � � ; T O � T � � � Apply ( : O � ) = : O � ^ = � ( � _ ) � ; T T � � � Apply & % 16 Hasan Da vulcu { Univ ersit y at Ston y Bro ok

Recommend


More recommend