Grid Resource Information Frameworks
Transcription
Grid Resource Information Frameworks
Frameworks. lnformation GridResource K. BoYanov K.KaloYanova, G.Pashov, GISdatadesign concerning someissues arelisted Below grid monitoring architecture, services, Key Words:Gridinformation presented theGIStechnological is Inthenextsection grid monitoring architecture features. relational modelis a keystepto Thechoiceof a suitableinformation Ahstract. lt hasto infrastructure. the Grid of use and deployment successful andperformance robustness for scalability, meetstrongrequirements and dynamicGridenvironment. in theheterogeneous models of theGridinformation tn thispaperthemaincharacteristics thathaveto be takenintoaccount: Thebasicchallenges arereviewed. complete functionally and replication, datadistribution architecture, Gridenti$ sets,etc.are adressed. representation for information of existingconcepts an overview Finally realizations of theparticular anda comparison in eiiA is presented andUDDIis provided. RDBMS basedonLDAP, services of information overview. by GIS to the DataProvided 2.1. MainRequirements provides about information isthatGls Themainpostulate real-world whose in GlS, Entities dataobjects. valuable abstract uniform clearand build objects, arethe representations gridresources. heterogeneous overthe layer ofcriteria mustmeeta number forGISentities Thecandidates [1]: - theentityincludes onlythevaluable . Usefulness information)' (only meaningful attributes - theentities from aredistinguishable . Uniqueness 1. Introduction entityamong oneparticular eachother(theusercanchoose of thesametYPe) allentities (GlS)is a complex - theentityis longlasting (transient Service . Persistence TheGridInformation meaningful and uniform rarely, structured, or extremely arenot thatmaintains accidentally architecture data,whichappears dynamic and heterogeneous in the aboutservices information forentities). candidates selection - the entityis valuable thediscovery, for multiple in orderto support . Generality Gridenvironment, diversity and the scale As resources. its of utilization optimal and or users. applications for needed or theinformation otherentities aregrowing, ofjoining haveto becapable oftheGridinfrastructure Entities requires crucial functionality becomes if their management category one efficient than more [1]' to belonging to theotherGridseruices. that.Hence relationships connected complex GISis closely mustsupport thedatamodel reliesontimelyaccess between in thegridenvironment Jobexecution entities. and for resources information and up-to-date to accurate Seruice Management Resource overwidearea. spread seruices, Kindof Datathat GISMustSupport site(0S,CPU, 2.2. operational mustbe ableto locatea suitable set. entity Grid complete functionally GISmustensure current aswellasto examine datasources) capacity, storage which of entities present variety whole the to means That storage available systemstate(CPUutilization, andexpected Forinstance above. outlined withthedatamodel comply job lt is scheduling. efficient to achieve in order etc.) space, jobprogress [ 1 , 6 ] : about information regular GISto provide important version, (operating system, . Static HostInformation with GISin orderto alsointeracts service too. Security etc.). processors, hostname, useroftheGISisthe Another users. andauthorize authenticate usage, CPU (load average, . Dynamic HostInformation is basedon the information whosefunctionality Gridportal, processes, etc.). running gathering byGlS. latency, (topology, bandwidth, . lnformation Network the happens, situation whenanabnormal lnparlicular, analysis etc,), inthesubsequent couldbeveryuseful information recent (totaldiskspace, . Storage Systemlnformation failure' the led to have that of theevents for reconstruction etc.). space, disk thesystem available datamaybeusedto optimize Thecollected etc')' services, . Applications (software, data,active onthefabric. to theconditions according behaviour roles,etc.). . Users(users, credentials, policies, etc.). (name, . 0rganisations security Modelsin Grid 2. lnformation . Instruments (radars, etc.). T h e i n f o r m a t i om n o d e li n t e g r a t easp p r o p r i a t e and a robust in orderto create anddatastructures technologies modelis very information GlS.Thechoiceof suitable reliable anduseof a Grid deployment for the successful imporlant infrastructure. 'Thispaper (Project n0. of excellence network by coreGRlD is supported (Project no.VU-Ml-110/2005)' SUGrid andproject FP6-00426) 2 2007 irrf orrnationteclrnologies anc|'corlEror It I 2.3.SomeConsiderations aboutData Dynamics Veryimportant aspectof Griddatamodelis data (which howit haschanged dynamics shows overtime). Such statistics couldbeusedfor:res0urce management; billing in no s ;s tl o a d e c o n o m igcr i d ;s y s t e mc r a s hi n v e s t i g a t i o h forecasting dataschema andso on.As a result, suppofting tracks forobjects' iscrucial. According changes tothedynamics, therearediscrete value's historical records for data(laststate, pastperiod)and continuous datastreams(thatreflect resource's of thedifferent ways dynamics onthefly),Despite it is desirable of usingdata,however, to havea uniform representation forexample mandatory timestamp thatincludes attribute, for Grid Information 3. Requirements Services Except for thedatamodel,the investigation of the is a characteristics, technologies andapproaches underlying keystepto successful deployment andefficient useoftheGrid infrastructure. There thathave to beexamined areseveral aspects in orderto acquire a clearnotion oftheGISarchitecture [1,2]. 3.2. Fault Tolerance events GISmustbe ableto copewithunexpected in a flexible whichcouldaffectresource availability manner (adaptation). Decentralized GISschema istherightapproach for it protects theGridfromtotalcrashincase thecasebecause However, in orderto ensure of failure of anyof its modules. a backup strategy morereliable andfaulttolerant environment, Forexample: a primary GISinstance stores mustbeavailable. orondiscandsynchronizes datawithstandby dataina memory would beusedincaseofcrash orforoptimization). GIS(which forGIS andjobstates) arecrucial Some of thedata(metadata onreserve sitealmost functioning andtheymustbesecured Therestpersistent historical) could immediately. data(mainly Themainaimis (fast)recovering withsomedelay. bedelivered fallenoffGISin shorltimemustbe fromabnormal situations: replaced withrelevant up-to-date GISinstance. andAdaptability 3.3"Reconfiguration itselfis a Theability of theplatform to self-organise forresilience Theplatform should vitalprerequisite to failures. givean opportunity of GIS for dynamic addition andremoval i n s t a n c eass a n a p p r o p r i arteea c t i otno c h a n g eisn t h e cond itions. environmental the ln supporting reconfiguration and adaptability, gathered plafform across mayutiliserulesandknowledge, runs. vs. Decentralised Approaches 3.1. Centralised A c e n t r a l i sa ep d p r o a c(hm a i n t a i n isnign g l ed a t a repository) isveryeasyto manage. Provider-centric GISensures thatthe platform is freedfromcomplex datamodelling and andCaching 3.4.Replication problems, lot binding issues but it introduces a of such as: [2] is replication Inthecontext of datagridtechnologies, . Performance: bottleneck lt couldbea pedormance a c c e s sl a t e n c ay n d b a n d w i d t h m o s t l yu s e dt o r e d u c e if all resources directtheirdatastreamsto a centralized arethatit Theotheradvantages of replication consumption. information server. availability bycreating helpsin loadbalancing andimproves . Scalability: grows,a Whenthenumber of clients precise information copies ofthesamedata.0btaining multiple singleinformation centremayhaveinadequate capacity to requirement fortheGISandreplication ontimeisanimportant process alltherequests, goodstrategies for achieving it. areextremely andcaching . Fault Whenthecentralized tolerance: server is not An Frequently needed datacanbelocally cached andreused. thewhole wouldstop.lt willberestored active system onlyafter question is howstale is.A mechanism important theinformation theinformation server is on line, TimeTo is necessary to guarantee data,,freshness". Providing (distributed A decentralized approach repositories), Live(TTL) attribute, indicating howmuchtimedatacanbeused, ontheotherhand,is muchmoredifficult to manage, involves is a goodchoice. communication andsynchronization butensures m0re overheads However, whenmultiple GISschemas' copiesare reliable fora GlS.Theconsiderations andflexible background provided? howis dataconsistency meta-data available, Static mentioned above arenolonger obstacles to goodGISrealization. is lesssensitive to changes synchronization delays andisolated Buttherearesomeothers: are not so critical.However, latencies in dynamic data . Distributed Search: In a decentralized environment result synchronization inreturning information, which in outdate meta-data is spreadovera largearea.To obtainrelevant turncouldleadto inefficient decisions. schedulinq information a number of repositories mustbe accessed. Distributed search means retrieving datafromspread information 3.5. Security centres andcombining resultsin fast and efficient way. problem. Ina Gridenvironment security is a complex Performance canbe significantly improved through indexes In order to address robust authentication and commu-nication whichcollect andaggregate frequently useddata. protection demands a suitable infrastructure security should be Anytopologies arepossible buta combination of the policy built. Adoption of too severe security could become twostrategies sounds morefeasible fora gridenvironment. We performance problem. lngeneral, GISschema contains resource partoftheactivities assume thatthegreatest is inthebounds jobs, characteristics and statuses of and current completed o f a d mi n i s t r a t i vdeo m a i n s( l o g i c acl l u s t e r a) n d i n t e r whicharenotstrictly confidential. So,balance between reliable communications areless,So,centralized information schema performance level security and adequate should be found. peer-to-peer on locallevelanddistributed infrastructure that providesglobal viewis themostreliable GISarchitecture. inf orrnatio5r andcontrol I t ec hnol ogrie s 2 2007 A schema contains the schema andregistry. othercomponents: (column ofeach types anddescription) names, name andstructure contains a list,foreachtable, A registry inthesystem. virtualtable data)forthattablei.e.it whoinsertrows(publish of producers infortabletheyproduce database to thelogical mapsproducers predicate usesSOLWHERE theproducer for.0n registration mation Thedataitself oftabledatait willpopulate. to definewhichpartition There statement. through SQLINSERT intologicaltable isinserted producer) (secondary thatcomcomponent is alsoanarchives producers and frommultiple information streams bines andfilters themin a database. archives Information of Existing 4. Overuiew query onthevirtual runSQLSELECT Theconsumers records, Thequery isf irst to fetchnecessary in order database Technologies whichproducers, for to identify the registry against checked query query, The is be the must contacted. virtualtable in each solution fora GIS mentioned, themostflexible Asalready producer, passed to obtain the answer relevant to each then thatmerges staticinformation and infrastructure usesa common theconrepository holding There is no central directly. forboth tuples a uniform schema monitoring dataandprovides dynamic virtual database. it is in this sense a the tables; tents of Architecture GridMonitoring andhistorical information. operational T h e m a i na d v a n t a goef t h e m o d e il s t h a t t h e gridinformation proposed (GMA), as Forum, models byGlobalGrid DBMS, whichis a proven and infrastructure relies onrelational (thatprovide (that information), consumers a setof producers writing there reading and and for both fast technology effective (integrated repository that request information) anda singleregistry products people DBMS familiar withrelational [6]. producers andconsumers). Fur- aremany mediates between theinteraction richsetof interfaces for dataextheGMAsupports thermore, publishisubscribe; notification 4 . 3 . U D D I andevent change: QUery/response Dle s c r i p t i oDni s c o v e a r yn d U D D It,h e U n i v e r s a [3,4,5,11,13]. provides which registry standard, is a Web Services Integration, to information therearetwomainapproaches Currently, There is also businesses services. fordescribing model a datamodel andrelational. Thehierarchical representation: hierarchical global public (UBR), which is a Business Registry the Universal intrees.Oneof theobjects is a rootof models objects asentries In addition UDDI the available services. registry for advertising to it. lt supports onlyonetypeof thetreeandtherestarerelated private registry However, community be up as a 0r can set [9]. parent-to-child (,,is-a"). Therelational modelreprerelationship: mostly forstaticdata. suitable sentstheobject typeas a relation; theobjects arethemselves it is a technology (MDS4) Discovery System The Globus' Monitoring and peer-to-peer (,,part-of") relationships tuples oftherelation. lt allows resources to monitor anddiscover is a suiteof webservices between relations. two WebServices on Grids.MDS4includes andseruices (WSRF)-based anIndex Service, Resource Framework seruices. 4.1. LDAP datafromvarious sources anda Trigger Seruice, thatgathers (LDAP) TheLightweight Directory Access Protocol is an which whenpre-defined trouble totakeaction canbeconfigured opennetworkprotocolfor querying andmodifying directory ser- conditions is a registry similar to aremet.TheIndexSeruice vicesrunning TheLDAP namespace is basedon overTCP/IP. WSRF but muchmoreflexible. Clients usestandard UDDI, property query entries whicharejoined inhierarchical Directory Information Tree. resource interfaces andsubscription/notification LDAP is designed for information retrieval ratherthanfor updating to retrieve froman Index. lndexes to information canregister information, andoffera staticviewof thedata. fashion inorder to aggregate eachotherin a hierarchical data provides provides TheGlobus Toolkit developers with atseveral application levels. Anadditional Archive Service access a hierarchical GISinfrastructure called Metadata Directory Service to historical data[7]. (MDS2) LDAP. using lt consists oftwocomponents: GridResource is oresented A shorlcomoarison of themethods on (GRIS) Information (GllS), thetable. Seruice andGridIndexInformation Seruice TheGRIS supplies information aboutprimary resources, while the GllSaggregates information frommultiple GRIS sources andpro5. Gonclusions videsa viftual scope forfastsearching 17,12,131. and 0pen SourceTools 3.6. OpenStandards andwellknown andprotocols Applying openstandards tools andbetter improved scalability toolshasasconsequence Thefundaexperience. development aswellassharing support GISinfraa flexible andreliable for building mental direction toolsand opensource be.openstandards, structure should portable (based Whenever openstanonJava). implementations it is better to specifications withgridproject dardortoolsagree usethem. 4.2. RelationalApproach presents (R-GMA) Relational GridMonitoring Architecture theinformation resources asavirtual database containing a setof tableswithcomplex relationships amongthem.NotethatR-GMA is notdistributed relational DBMS butrather a wayto userelational principals in distributed (consistency environment requirement is notsoemphasized in somecases) [6,8,10,11,131. producers provides Besides and consumers, R-GMA two T h e s u c c e s s f ud le p l o y m e a nn t du s eo f t h e G r i d infrastructure strongly depends on the choiceof a suitable information model. Theabove review information oftheexisting technologies canhelpinthataim.Thehierarchical datamodel grid iswellsuited to thedistribution needs ofthecomputational because the rooted treelendsitselfwellto partitioning into byadministrative subtrees domains. UDDI, asthenewest and promising tendency in theGridinformation infrastructure, has 2 2007 irrf orrnation technol ocrie s andcoritrol ,d LDAP Mvantages protocol. Openstandard,widespread Extensible information model Directoriescan be distributedand to providescalability and replicated reliability.High levelof dataprotection and data access. Standardlnternetprotocolsfor data access. Disadvan- Poorupdateperformance. tages A huge amountof data can not be efficiently managed. For unusualquestions,gettingan answercouldbe very expensive. The LDAPquerylanguageis limited (for examplethere is no join operation). Radicalchangesin the informational modelare not possible. RelationalModel UDDI Extendable model(additional O p e n a n d i n c r e a s i n g l y canbe addedvery information Flexible supported standard. are manypeople and scalable. easily).There DBMS familiarwithrelational products. RelationalDBMSare basedon strictACID principles,but in GIS this strongconditions could lead to performance problems. Some of the systemsare too c o m p l e xa n d t o o h a r d f o r administration. The usingof data partitioning for a securitypolicyis a more d i f f i c u l tt a s k f o r m u l t i p l e domains. administrative platform forGlS.However, from a potentialto become a steady largely thepresented technologies therelational approach covers platform theGISrequirements andensures themostsuitable f o r b u i l d i n rge l i a b laen d r o b u s tG I Sb a s e do n o p t i m a l fordatamanipulating. environment platform, whenchoosing theproper some Certainly, likefaulttolerance, otherconsiderations datadistribution, replication, andconsistency areessential as well. D i f f i c u l t i e si n h a n d l i n g dynamicinformationthat requiresf requentupdating. No explicitdatatypingfor informationin the UDDI directory. Limitedquerylanguage Evaluation as a M., Humphrey. of UDDI E.,G.,Wasson, 9. Benson, for 0GSA-based Grids.April Provider Discovery Services of Resource 2006. M. Ruda.GridInfrastructure 10. Holub, P.,M. Kuba,L. Matyska, AcrossGrids Information European asReliable Service. Second Monitoring 2004. Conference, M.0evers, P. A.,A. Gray, Ma 1.,W. Nutt,J. Magowan, 1'1.Cooke, M. Soni,A. Wilson, R. R. Byrom, L. Field, S. Hicks, J, Leake, Taylor, B.Coghlan, A.Djaoui, N.Podhorszki, L.Comwall, S.Fisher, Cordenonsi, Integration R-GMA: AnInformation System S: KennyandD.0'Callaghan. lCClS, 2003. for GridMonitoring. M. Mirto. L Epicoco, S. Fiore, 12. Aloisio, G.,E. Blasi,M. Cafaro, References Across Information Service. FirstEuropean Grids Dynamic GridCatalog 2003. Conference, 1. Plale, 8., P.Dinda, G.von Laszewski. KeyConcepts andServices of M.,R.,Sakellariou. Information Services for Large-Scale 13.Dikaiakos, a GridInformation ICDCS 2002. Service. Engine. Technical Report, A Casefor a GridSearch CoreGRlD 2. Badia, R.,0, Beckman, M.Bubak, D.Caromel, V. Getov, S. lsaiadis, Grids: 2005. V. Lazarov, M.Malawski, S. Panagiotidi J. Ihiyagalingam. Lightweight Design GridPlatform: Methodology. October 2005. Manuscript received on11.06.2007 3. Zhang, X.,J.Freschl, A Performance J.Schopf. Studyof Monitoring andlnformation Services for Distributed Systems. HPDC-12, 2003. 4. Byrom, R.,B. Coghlan, R. Cordenonsi, A. Cooke, L. Cornwall , M. GeorgiPashovreceived hisMasterof Science Craig,,A.Djaoui, S. Fisher,A. Gray,S. Hicks, Kenny, J. Leake, 0. degree from the in 1997.Since Sofia University Lyttleton, J. Magowan, R. Middleton, W. Nutt,D. 0'Callaghan, N. 2005 he has beena PhDstudentat Sofia Podhorszki, P. Taylor, J. Walk,A. Wilson.Production for Services Hisresearchinterests University. are in the Information andMonitoring in theGrid.UKe-Science Conf.,September fieldsof databases, data-warehousing, GRID 2004. computingand informationsystems. B.,A.Djaoui, 5. Coghlan, S. Fisher, J. Magowan, M. 0evers. Time, Information Seruices andtheGrid.May2001. CpfracE 6. Dinda, P.,B.Plale. A Unified Relational Approach to GridInformation lnstitutefor ParallelProcessing, Bulgarian Services. February 2001. Academy of Sciences 7. Aloisio, G.,M. Cafaro, l. Epicoco, S. Fiore.Analysis of theGlobus Tel:+ 359 2 9215346, ToolkitGridInformation Seruice. Version 1. gpashovoacad.bg e-mait 8. Allan, R.,D.Chohan, X. Dong Wang,A.Richards, M. McKeown, J. Colgrave, M.Dovey, M. Baker, S. Fisher. Buibing tlree-Science Gridin tobecontinued onpage16 theUK:GridInformation Services. UKe-Science Conf., September 2003. irrf grrnatio5r andcontrol teclrnologies 2 ?007 References 13. Landauer, R. lrreversibility andHeatGeneration in theComputing Process. IBMJournal, 196'1, 183-191. (htp://uamru, resqarch.i bm,c-om/jo_u 53libmrd0503C.pdf). rnal/r:d/0 1. Pavlov, V. Efficient Quantum Computing Simulation.proceedings 14. Eisert, of J. andM. M. Wotf . 1uantum Computing. I E E EJ o h nV i n c e nAt t a n a s o 2f f0 0 6 i n t e r n a t i o nSayl m p o i i u m (http;xxx. phi04010l 9). lanL.qovlp-dflquanl on Modern C o m p u t i nIgE. E EC o m p u t eSro c i e t py r e s s 2, 0 0 6 , 15. DiVincenzo, D. Two-bitGates arelJniversal for luantum 235-239. Computation (http//xxx,larl.gov/pdfI quant-ph/ 940l 022). 2. Shor,P.Algorithms forOuantum Computation: Discrete Logarithms 16. Barenco, A. A Universal Two-bit Gatefor Quantum andFactoring. Proceedings of the 35thAnnual Symposium on Computation (htp_.lxxxJ an!.govlp.d flquart:phl950501_0 ). press, Foundations of Computer Science. IEEE Computer Society 17. Deutsch, D.,A. Barenco andA. Ekert.Ilniversality in luantum 1994,124-134. (hfiuiAu-Lanlg_oVpdllq uant-ph/950j 0?L). Computation (htp:lxxx,tanl.go,vlpdf& uanLph/950501B). 3. Barenco, A., C. Bennett, R. Cleve, et al. Elementary Gates for 18. Deutsch, D. andA. Ekert.Machines, Logicandluanfum Quantum Computation. Physical ReviewLettersA, S2,1995,3457. Physics (httB:Au-Lan! gsv/pdVmattr.H0,1 99I .1t 50). 4. V edr al, V ., A. Ba re n caon d A , Ek e rt.0 u a n tumN etw orks for Elementary Arithmetic 0perations. (lrttp//xxx.tanl.sov/pdf/q uant:phl95 I 101_8) 5. Benett, C. LogicalReversibility of Computation-tBM Journalof Besearch andDevelopment, 1973,17,525-532. (htp;//wwwresearch.ibm.e,0mlio_u rxaltd/l 76 ib-nrr 7-06G.pdl). d_1 Manuscript received on12.06.2007 6. Dirac,P. A. M. ThePrinciples of Ouantum Mechanics. OxfordUniversity Press, Oxford, 1947. Eng.ValentinPavlov(born1974) hasgraduated from the TechnicalUniversityof Sofia during 1997, ,pomputingSystems"speciality, with MSc thesis,,Algorithm for discovering the generators,orbits and the order of 7. Lens t r a, H., H . L e n s traJ ,r., M. M a n a s s e t, a l . TheN umber group of a graph". Since the automorphism FieldSieve. Proceedings of the 22ndAnnual ACMSymposium on 1999 he is workingin Rila SolutionsEAD, Theory of Computing, ACM,NewYork,1990,564-572. currently occupying the Chief Technology 8. Pavlov, V., H. Turlakov and K, Boyanov. Quantum Computation ]fficer position, Since 2002 he is a PhD and O uant u m (Ko a H ro Bo C o mp u te rs i l s q h c re H ue i l K B aH ToB t4 student at the lnstitde for ParallelProcessing xoNnnrpu).-Engineering Sciences, XLll,2005,2, S-28. at he Bulgarian Academyof Sciences. ln July proceedings 9. Deutsch, D. Quantum Computational Networks. of 2007 he passed the preliminarydefenseof his phD thesis,,luantum theRoyalSociety, Ser.A, 425,1989,73-90. l0.Benioff as a physical , P. TheComputer System: Microscopic ComputingSimulation".He is the laureateof the 2005 ,,JohnAtanasoff" Quantum Mechanical Hamiltonian Model of computers asRepresented Award of the Presidentof Republicof Bulgariafor scientificand prophysics, fessianalachievements by TuringMachines.-Journal in he lT field. His scientific interestsinclude: of Statistical 22, 1990, quantumcomputing,GRID,distributedsystems,computerand network 563- 59 1. 11.Feynman, R. Simulating Physics with Computers liternational architectures,protocols and standards. Additional Sources Journal of Theoretical Physics, 21, 1982,46-488. 12.Deutsch, D. andR. Jozsa.RapidSolution of problems by Quantum Computation.-Proceedings of the RoyalSociety, Ser.A, pp .5 5 3 -5 5 8 . 439,1992, Contacts: e.-m g ail: upaulovoila.b continuation trom5 Prot.DScKiril Boyanov is a Academician at the Bulgaian Academy of Sciences. He is Director of KalinkaM. Kaloyanova received herM.Sc.in lnstitutefor Parallel Processing.He is theButgan Computer Sciencefrom llniversityof Sofia ianRepresentative in lFlP;Chairman of Bulgarian (1981),and her PhDdegreein Computer Computer Societyof IEEE; Chairman of National SciencefromtheCentral lnstituteof Computer lFlPCommittee; NationalCoordinator of IST Technique and Technologies, Sofia(lg|g).She Programme; Memberof ISTAdvisoryGroup wasResearch Fellowin theCentrallnstitute for - European (ISTAG) Commission. He hasmore Computing Technics and Technologies, Sofia publications than 100 in Bulgaian,morethan (1981-1995) andBTC lnstitute for Scientific 40 publications in foreignscientificmagailnes Researc h of Telecommu nications ( | ggS-200il. and internationalconferences proceedings. since2003she shehasbeenworkingas Assoc.professor at the Afihor and co-authorof 36 books and manuals(in Butgarian mainly, University of Sofia,Facultyof Mathematics andlnformatics. severalof hem publishedin Russian). Editorof 4 booksof conference proceedingspublishedby North-Holland.chief Editor of journal ,,lnformation Technologies and Control"- Bulgaria. eontacls: e-mail: kkaloyanovaofmi-uni-sofia.bg ItJ 2 2007 Contacts-: e-mail: boya novoac ad.bg i nf orrnati on t ec trnolocries andcon:[rol