Performance Challenges in Object-Relational DBMSs.

摘要: XML is rapidly becoming a popular data format. It can be expected that soon large volumes of will exist. either produced manually (like html documents today), or it generated by new generation software tools for the WWW and/or electronic interchange (EDI). The purpose this paper to present results an initial study about storing and querying data. As first step, was focussed on use relational database systems very simplistic schemes store query In other words, we would like how simplest most obvious approaches perform, before thinking more sophisticated approaches. general, numerous different options addition database, stored in file system, object-oriented (e.g., Excelon), special-purpose (or semi-structured) system such as Lore (Stanford), Lotus Notes, Tamino (Software AG). still unclear which these ultimately find wide-spread acceptance. A could used with little effort data, but not provide any support Object-oriented allow cluster elements sub-elements; feature might useful certain applications, current mature enough process complex queries databases. going take even longer are mature. Even when using RDBMS, there many ways One strategy ask user administrator order decide tables. Such approach supported, e.g., Oracle 8i. Another option infer from DTDs should mapped into tables; has been studied [4]. Yet another analyze workload; devised, [2]. work, only simple ad-hoc schemes; think necessary adopting approach. require no input user, they work absence if meaningless, do involve analysis Due their simplicity, show best possible performance, see, some them good performance situations. Also, guarantee known so far perform better than our see [3] experimental respect. Furthermore,