CMSC724: What goes around comes around

Size: px

Start display at page:

Download "CMSC724: What goes around comes around"

Willis Greer
6 years ago
Views:

1 comes around University of Maryland, College Park January 31, 2012

2 Data Modeling A data model theory typically involves structural part collection of concepts/constructs to represent data integrity part constraints to ensure data integrity manipulation part constructs for manipulating the data We would like it to be: Sufficiently expressive can capture real-world data well Easy to use Lends itself to good performance

3 Hierarchical/IMS Constructs: Hierarchy, keys, record types, DL/1 lang. Issues: Navigational interaction No physical data independence Repetition of information (m-to-n relationships) Inability to represent information Last two solved by a latter extension Some logical data independence

4 Network/CODASYL Constructs: set type, network structure programmer as a navigator Cons: No physical or logical data independence Bulk loading 3-way relationships Very complex programming constructs

5 dix A Network Model Data Models: Network/CODASYL currency pointers Turner Putnam Stamford customer template A account template Round Hill Horseneck branch template customer account branch depositor account_branch run unit Hayes Main Harrison We have enclosed part of the query in a while loop, because we do customer.customer city := Harrison ; Johnson Alma Palo Alto find any customer using customer city; while DB-status =0do Turner Putnam Stamford begin get customer; Perryridge Horseneck print (customer.customer name); find duplicate customer using customer city; end; Downtown Brooklyn Round Hill Horseneck advance how many such customers exist. We exit from the loop when This action indicates that the most recent find duplicate operation fa that we have exhausted all customers residing in Harrison. A A A A Appendix A Network Model A.4.4 Access of Records within a Set A The previous find commands located any database record of type <re A-201 this900 subsection, we concentrate on find commands that locate records DBTG set. The set in question is the one that is pointed to by the <set-ty pointer. There are three different types of commands. The basic find co find first <record type> within <set-type> Figure A.20 Program work area. which locates the first member record of type <record type> belong rent occurrence of <set-type>. The various ways in which a set can b discussed in Section A.6.6. get, which copies the record to which the current of run-unit points from the database to the appropriate program work area template To step through the other members of type <record type> belong occurrence, we repeatedly execute the following command:

6 Relational Constructs: Relations, relational algebra/calculus, functional dependencies Physical and logical data independence Cons: Transitive closure (initially) performance (initially) too complex and mathematical languages

7 Don Chamberlin of IBM was an early CODASYL advocate (later co-invented SQL) He (Codd) gave a seminar and a lot of us went to listen to him. This was as I say a revelation for me because Codd had a bunch of queries that were fairly complicated queries and since I d been studying CODASYL, I could imagine how those queries would have been represented in CODASYL by programs that were five pages long that would navigate through this labyrinth of pointers and stuff. Codd would sort of write them down as one-liners. These would be queries like, "Find the employees who earn more than their managers." [laughter] He just whacked them out and you could sort of read them, and they weren t complicated at all, and I said, "Wow." This was kind of a conversion experience for me, that I understood what the relational thing was about after that.

8 scriptions error. for nondata-base functions such used as a mnemonic device for the status Frank as input/output and computation. For those ations. The code can examine DATABASE- In the event that an error code results condition of a data-base. It is initialized Network who know the Data COBOL Model: programming Example lan- STATUS as well as other status locations in from a DML operation, control is returned to the correct Program status value. While using goesa guage, The the corresponding IDENTIFICATION code should and be EN- an attempt to determine what comes caused aroundthe to that section within the DECLARATIVES DBTG-like system, it might be common obvious. VIRONMENT DIVISIONS remain un- error and how the program should attempt that corresponds to the error code. From article in: ACM Computing Surveys, 1976 Our COBoL/DML program begins with practice For changed from those used by standard to to recover define from a library it. of common every error status code listed explicitly the standard COBOL IDENTIFICATION status in a COBOL, Within the ENVIRONJvIENT DI- codes Following and to the include DECLARATIVES them in the section USE FOR DATABASE-EXCEPTION and ENVIRONMENT VISION, we assign DIVISIONS: an internal COBOL file working ON are storage the section normal by using processing the COBOL procedures: "error status code", control is passed to to COPY the facility. NO-MORE-SONS and NO- IDENTIFICATION the actual print DIVISION. file, for output of our INITIALIZATION. paragraph that follows the USE statement. query. PROGRAM-NAME. SAMPLE-QUERY, READY PRESIDENTIAL.AI~A. MORE-STATES are status variables used ENVIRONMENT DIVISION. OPEN nen-databaee COBOL fil~. In the preceding example, if a DATABASE- The identification COBOL of machine DATA environment DIVISION and incorpo- within the program MOVE "FALSE" logic. TO NO~-31OltE-STATES. STATUS d~ acafion of non-data-base of "04021" files were returned, the FIND FIRST STATE IN ALL4~TATES.$S. rates (i.e., the atandatd link COBOL to the filos) data base: The procedures PERFORM for accessing PROCESS-STATE the data TIIIRU base FINISH-STATE paragraph labeled EXPECTED-ERROR- UNTIL NO-MORE.STAT~ - "TRUE". DATA DIVISION. GO TO FINISH-UP. HANDLING would be invoked. Any FILE SEC~rIoN. PROCFESS~STATE. DB PRES.ADMIN-STATE-INFO WITHIN PRESIDENTIAL. elimi- DATABASE-STATUS codes that are not MOVE 0 TO PRESIDENT-coUNT. Computing FD Surveys, REPORT-FILE. Vol. 8, No. 1, March 1976 IF NATIVE-SON IS EMPTY e subexplicitly remainder listed of data item cause entries which a transfer make to the para- MOVE "TRUE" TO NO-MORE.SONS, up a standard COBOL file. ELSE MOVE "FALSE ~ TO I~,'O-MORE.SOI~;S. items graph following the USE FOR DATABASE- WORKING-STORAGE SECTION. PERFORM COUNT-NATIVE-SOI~S 77 PRESIDENT-COUNT USAGE COMPUTATIONAL PIC 999. using EXCEPTION ON OTHER statement. In UNTIL NO-MORE-SONS - "TRUE" DONE PIC 9(5) VALUE "04021". GO TO FINISH,STA'I~ the 77 NO-MORE-SONS preceding PICA(5). example, control would be COUNT-NATIVE-SONS. 77 NO-MORE-STATES PICA(5). FIND NEXT PRESIDEET IN NATIVE-SON. are returned specified to in the the paragraph COBOL PROCEDURE UNEXPECTED- an expected IF error DATABASESTATUS situ~tlon occurs:when -- DONE we DIVISION: ERROR-HANDLING The DB entry specifie,~ which after sub-schema an unlisted sequentially MOVE traverse "TRUE":thr~l.Jl TO K0-MORE.80!NS a st oc- ELSE ADD 1 TO PRI~IDENT-COUR'T. re de- and DATABASE-STATUS PROCEDURE schema DIVISION, this program statement is referencing. resulted. currence and FINISH.STATE. reach the end of the set oc- DECLARATIVES. IF PRESIDENT-COU~,~T IS GIR.I~ATER THAN 1 ritten While. EXPECTED-ERROR As implied not required by SECTION. the by paragraph the specifications, and section currence. Such FIND an STATE error CURRFE~'T, situation usually USE FOR DATABASE-EXCEPTION ON "04021". speci- the EXPECTED-ERROR-HANDLING. effect of such a statement in most im- means that GET we STATE, names in the preceding example, there is should ~ proceed to another write out state rise sn4 I ~ ~omllt, ta-base plementations generally EXIT, a distinction to cause between the record error de- situ- part of our FIND program NEXT STATE to~ continue IN AL]~TATES.,~I8 processing. UNEXPECTED-ERROR SECTION. IF DATABASE.STATUS' = DO~E e first scriptions ations USE FOR that from DATABASE-EXCEPTION we the expect sub-schema to happen ON OTHER. (augmented as part of An example MOVE of an "TRUE" unexpected TO NO-MORE-STATES. error condi- UNEXPECTED-ERROR-HANDLING. ta-base by our schema here normal we would information) processing, procc.~s unexpected to error and be copied error into situations the tion is an FINISH-UP. input/output error (for example, FINISH PRESIDENTIAL.AREA. assume COBOL conditions. application program, thus reserving bad parity detected). CLOSE non.datalmee COBOL filee. that END are DECLARATIVES, totally unexpected. An example of STOP RUN. mples. space within the program for each data-base In this example the DATABASEfinds record The first type (and part selected of the items) PROCEDURE which this STATUS code of "04021" corresponds to an one DIVISION program may is the access. DECLARATIVES The record and sec- data the end-of-set occurrence condition just mentioned. When suchl an error occurs, the

9 Entity-relational Constructs: entity, relationship Limitations: Lack of query language, ease of mapping into relational The conceptual model of choice to design the schema Relational++ Constructs: set-valued attributes, aggregation, generalization etc Limitations: Didn t prove terribly useful either functionally or performance-wise

10 Semantic data models OO Constructs: class, class variable, multiple inheritance etc Essentially persistent PLs Less impedance mismatch Weak support for Xions, queries etc. Niche market (e.g. CAD)

11 OR Constructs: User-defined functions, types, access methods Semi-structured Schema last Constructs: DTD, XMLSchema, union types etc Issues: Limited applicability??, doesn t really solve semantic heterogeniety Standard for wire format

12 Some lessons Physical/logical data independence desirable Record-at-a-time, navigational interfaces force manual query optimization and won t scale Technical debates settled by the marketplace in many cases KISS OO: Packages will not sell to users without major pain User-defined functions and access method effective Schema-last is probably a niche market Semantic heterogeneity not solved by XML

CMSC724: Data Models

CMSC724: Data Models University of Maryland, College Park January 29, 2013 Outline 1 Databases: A Brief History 2 What goes around comes around Databases: A Brief History 1960 s: Computers finally become attractive, and enterprises