NYC Wed Jan 22 meeting Informix

Title: Object-Relational Database (Informix/Illustra) Speaker: Donald Payne, Informix Software, Inc Date: Wednesday January 22, 1997 Time: 6:15-9pm Location: IBM, 9th floor, 590 Madison Av (57th St), NY NY Group: Database SIG of Object Developers Group, Inc (ODG) see http://www.objdev.org Leader: Des Shaw, "dshaw@ilx.com" RSVP: Appreciated, to: "dshaw@ilx.com", with your name, email, organization and title (if any). Cost: Free. All are welcome to attend. Audience: This meeting has two parts: a brief intro to O-O and O-R databases, followed by a detailed presentation on the Informix/Illustra OR-DBMS, with examples. It will be of greatest value to attendees interested in O-O, O-R and DBMS concepts, who develop programs that call or use a DBMS via SQL or an API. Agenda: 6:15-7pm Intro/tutorial on O-O and O-R DBMS 7-9pm Informix OR-DBMS Capabilities * Brief "Why OR-DBMS?" comparison of pre-R-DBMS, R-DBMS, OO-DBMS, OR-DBMS * Brief company overview (Illustra, Informix) * Illustra & INFORMIX-Universal Server capabilities * create type: opaque (base), row (composite), constructed (SetOf()...) * create function: strongly typed * operating binding to function name * sub-types & sub-tables; inheritance * public and private attributes & methods (Grant, Revoke on columns, functions) * overloaded functions; late binding * aggregate functions * functional indices (index on computed value) * virtual columns (compute on select) * "C" API * new access methods (indices) * Virtual Table Interface (VTI) in IUS (aka Extended Access Methods Interface EAMI in Illustra) * DataBlade modules * Examples (will depend on available time) - Bill of Materials explosion - Time Series query aggregate functions, e.g. Trailing average, correlation between 2 time series - Text search - Web Blade (not object-oriented, but handy extension) Informix: Informix Software, Inc. focuses on these areas: * Database management systems (DBMS) * Development tools * Connectivity software Informix's main DBMS's are: (see http://www.informix.com) * OnLine Dynamic Server family of Relational DBMS's * Illustra(tm) Object-Relational DBMS (OR-DBMS) and related DataBlade(r) modules (pre-built extensions) * Informix(r)-Universal Server which is an O-R DBMS, and the result of the code merger with Illustra. Has compatible Datablade extensions. DataBlade modules are prebuilt extensions (from Informix or 3rd parties) that support new complex data: types, functions, and possibly access methods. Examples: full- text search, spatial data, image matching, time series. Bio: Donald E Payne is a Senior Systems Engineer at Informix, where he performs pre- and post-sales support and consulting. In a recent project, he built a Web site based on the Illustra OR-DBMS for Electronic Buyer's News (EBN), a CMP Media magazine. Donald joined Informix when it acquired Illustra Informa- tion Technologies, Inc.. At Illustra, Donald trained users in addition to his consulting function. Doorprizes: (1 copy) software: "OnLine Workstation Server", a Windows-based single-user product based on Informix DSA (Dynamic Scalable Architecture) (2 copies) book: "Object-Relational DBMS - The Next Wave", Michael Stonebraker (Morgan Kaufmann Publishers, 1996) Comments: An OR-DBMS is a relational DBMS that is extensible in its types, functions (methods), and access methods (indices). It also has some O-O features, such as type constructors (eg, setof(type)), type and table inheritance, and overloaded functions. The extensions can be accessed by an enhanced SQL ("SQL 3"). The advantages include reducing program development time by a good "impedance" match between the business world and the data model, and also between the software structure (C or C++) and the data access methods. OR-DBMS incorporate many O-O features, while retaining the decade long proven technology of relational databases. Legacy code using conventional Relational SQL can comfortably co-exist alongside the newer O-O programs, and use the same databases. This facilitates evolution in code design, rather than requiring a sharp break when moving to O-O tools. In practice, the efficiency of an OR-DBMS approach may be about the same as a "pure" OO-DBMS (this is open to dispute :-). The constraining factors are more likely to be the database design, data storage decisions and the inherently serial O-O processing. OR-DBMS and OO-DBMS are each well suited to their appropriate problems. For some applications, which can be handled either way, the O-R approach may have a practical advantage due to the installed base of programs, and of people educated in relational systems. Additional info on OR-DBMS can be found on their website: http://www.informix.com/informix/corpinfo/zines/whitpprs (C) Copyright 1997 by Object Developers Group, Inc. All Rights Reserved.