| OCR Text |
Show CHAPTER 1 INTRODUCTION rrhe database needs of toclay's application s, especia lly in la rge compu ter-aided des ign and e ngineerin g ( Cf\ D /CAE), di st ributed cooperat ive computing, cornpu ter< t.ided softw are e nginee ring (CASE), <-r nd of6ce informat ion syst ems, a re very diffe r<' UL from tbose of t he bus in ess problems of' the 1070's when t he relat ion a l database model was introduced. l\!Iost of toclay's applications typically require creation, sLo rag<', a nd m<tn ipulation of la rge graph-based structures. Such appli cations need hoth associat ive retri eval (i. e ., query) and graph traversal (i .e. , navigation) . They gene ra lly proceed in dist inct phases of loading large working sets, working , and saving. T he relat io na l model was not des igned for such appli cat ions. Moreover, it was uot d<"s ig rwd to hc-t.nclle cl<t.La typ <·s like voice ., vi cl f~o, text, a nd irn ages . O bjectorit- · rrl<'d cl a.tahases (OODBs) [6 , 10 , 14 , 21, :3] a n~ designed to br ing object-oriented technology into a DBMS and solve some of these problems. One of Lhe main components of an OODB is a P e rsistent Object Store (POS). POSs support the creation, manipulat ion , storage, a nd ret rieval of o bj ects. Ot her dPsi rab l<-' features of pers isteut sto res a re protection , t ransaction management, version cont rol, and co nc.urr<c~ n cy control. T he re is a. great deal of vari at ion in t<~ rnls of the a rn ounL of funct iona li ty supported by different POSs. One reason for this is th at the dema nds of appli cation s a re not well unde rstood. Because of t his lack of knowledge, systems like Mneme [17] p rov ide mecha nisms, not poli cies, leaving t hese to be t uned, controlled, and extended by t he a pplication programmer. One |
