`RPX v. AIT
`
`
`
`
`OCTOBER 21. 1985
`.
`9-
`COMPUTERWORLD
`
`IN DEPTH
`
`This material may be protected by Copyright law (Title 17 U.S. Code)
`(cid:3) (cid:55)(cid:75)(cid:76)(cid:86)(cid:3)(cid:80)(cid:68)(cid:87)(cid:72)(cid:85)(cid:76)(cid:68)(cid:79)(cid:3)(cid:80)(cid:68)(cid:92)(cid:3)(cid:69)(cid:72)(cid:3)(cid:83)(cid:85)(cid:82)(cid:87)(cid:72)(cid:70)(cid:87)(cid:72)(cid:71)(cid:3)(cid:69)(cid:92)(cid:3)(cid:38)(cid:82)(cid:83)(cid:92)(cid:85)(cid:76)(cid:74)(cid:75)(cid:87)(cid:3)(cid:79)(cid:68)(cid:90)(cid:3)(cid:11)(cid:55)(cid:76)(cid:87)(cid:79)(cid:72)(cid:3)(cid:20)(cid:26)(cid:3)(cid:56)(cid:17)(cid:54)(cid:17)(cid:3)(cid:38)(cid:82)(cid:71)(cid:72)(cid:12)
`
`Does your DBMS
`run by the rules?
`
`
`
`To be ‘3mid+ 805”fully relational, a
`DBMS must support all 12 basic rules
`plus nine structural, 18 manipulative
`and all three integrit
`rules. "There
`will be more requirements by the;1990s.y'
`
`By E. F. Codd
`
`Last week’ the originator ofthe Noexisting DBMS product that I know ofcan honestlyclaim
`
`.
`,
`relational model described the 12
`rules by which to measure any
`.
`.
`DBMS Claiming to be relational
`This week, Dr. E. F. Codd presents
`-
`-
`the practical consequences of his I2
`rules 618 well as 30 (lddifiOflfll -
`features of“ relational Syste,n_ Then
`be asks vendors to measure up.
`
`Part 2
`
`to be fully relational at this time. The proposed ANSI
`standard does not fully comply with the relational model, so
`a DBMS’ fidelity to the ANSI standard is no guarantee of relational
`capability. The standard could be modified, but already vendors are
`well advised to extend their products beyond the standard to support
`customers’ DBMS needs fully.
`In their ads and manuals, vendors have translated the term
`lly relational" to ”fully relational," so more stringent criteria
`must be applied. Twelve rules (below) comprise a test to determine
`whether a product that is claimed to be fully relational is actually so.
`A grading scheme used to measure the degree of fidelity to the
`,e1_.,fi,,n,.,1m.,de1 fo1]ow5_
`A DBMS advertised as relational should comply with the follow-
`’
`12 rules:
`mg . The information rule,
`The guaranteed access rule.
`Systematic treatment of null values.
`Active on-line catalog based on the relational model.
`The comprehensive data sublanguage rule.
`The view updating rule.
`High-level insert, update and delete.
`Physical data independence.
`. Logical data independence.
`10. Integrity independence.
`1 1. Distribution independence.
`12. The nonsubversion rule.
`
`
`‘°9°.‘~'.°"5":“§*’."""
`
`E. F. Codd originated the relational model for data base management.
`He led the team that designed and implemented the first operating
`system with multiprogramming capability. This year he established two
`companies with Chris Date: The Relational Institute and the Cadd Er
`Date Consulting Group, both based in San lose, Calif.
`
`
`
`50
`COMPUTERWORLD
`OCTOBER 21, 1985
`
`IN Iii!’-‘I'll/RELATIONAL DBMS
`
`age are applied, holds con-
`siderable interest for users.
`The Oracle DBMS in particu-
`lar has an outstanding ap-
`proach to null values. The
`user may specify whether
`the aggregate function is to
`ignore null values or yield a
`null result if any null value
`is encountered.
`In general, controversy
`still surrounds the problem
`of missing and inapplicable
`information in data bases. It
`seems to me that those who
`complain loudly about the
`complexities of manipulating
`nulls are overlooking the
`fact that handling missing
`and inapplicable information
`is inherently complicated.
`Going back to progra.mmer-
`specified default values does
`not solve the problem.
`Rule 6, the comprehensive
`
`
`The ANSI
`standard as
`now proposed is
`quite weak. It
`fails to support
`numerous
`
`features users
`need to reap the
`advantages of
`the relational
`approach.
`
`data sublanguage rule, is im-
`portant for several reasons.
`First, it allows programmers
`to debug their data base
`statements interactively,
`treating them separately
`from whatever nondata base
`statements occur in their
`programs — a significant
`contributor to productivity.
`Second, it means that a sin-
`gle tool can be used for defin-
`ing relations derived from
`the data base, whatever the
`purpose. The view updating
`rule, Rule 6, is vital for the
`system to support logical
`data independence.
`Rule 7, which requires a
`multiple-record-at-a-time at-
`tack on insertion, update and
`deletion, can help save a
`good portion of the total cost
`of intersite communication
`in a distributed data base. If
`the system includes a good
`optimizer (an important com-
`ponent in relational DBMS
`performance), this rule can
`also result in substantial sav-
`ing of CPU and 1/0 time,
`whether the data base is dis-
`tributed or not.
`Failure to support inde-
`pendence (Rules 8 through
`11) can, and very likely will,
`result in skyrocketing costs
`in both money and time. De-
`veloping and maintaining ap-
`plications programs and ter-
`minal activities will be more
`expensive. Managers may
`even be unwilling to consider
`changing certain business
`policies simply because of
`the anticipated program
`maintenance costs.
`Rule 12, the nonsubver-
`sion rule, is crucial in pro-
`tecting the integrity of
`
`_These rules are based on a
`single foundation rule. I call
`it Rule Zero:
`For any system that is ad-
`vertised as, or claimed to be,
`a relational data base man-
`agement system, that system
`must be able to manage data
`bases entirely through its re-
`lational capabilities.
`This rule must hold
`whether or not the system
`supports any nonrelational
`capabilities of managing
`data. Any DBMS that does
`not satisfy this Rule Zero is
`
`not worth rating as a rela-
`tional DBMS.
`But compliance with Rule
`Zero is not enough. Failure to
`support the information rule,
`guaranteed access rule, sys-
`tematic nulls rule and cata-
`log rule can make integrity
`impossible to maintain.
`These four rules support sig-
`nificantly higher standards
`for data base administration
`and control (authorization
`and integrity control) than
`earlier DBMS supported. Us-
`ers should remember that a
`
`data base managed by a rela-
`tional DBMS is likely to have
`both experienced and inex-
`perienced users; it must be
`able to serve both.
`
`Rule Zero not enough
`Rules 1 and 4, the infor-
`mation and catalog rules, al-
`low people with appropriate
`authorization (such as exec-
`utives of the company) to
`find out easily via terminal
`what information is stored in
`a data base. I have encoun-
`tered data base administra-
`
`tors using nonrelational sys-
`tems who were unable to
`determine if a specific kind
`of information was recorded
`in their data base.
`Rule 3, which calls for the
`inclusion of systematic sup-
`port for unknown and inap-
`plicable information by
`means of null values that are
`independent of data type,
`should help users to avoid
`foolish and possibly costly
`mistakes. The treatment of
`nulls, when aggregate func-
`tions such as total and aver-
`
`2
`
`. .to mn . ..
`
`- Our new UPS
`is indeed a
`perfect match for lBM. ll
`.
`provides the kind of power
`protection necessary for IBM systems. And only Elgar is
`plug-in compatible with both the CPU and peripherals
`for fast. easy installation.
`- Our new UPS is packed with quality features for troublefree
`performance. Only Elgar has 14% dynamic power regulation
`We deliver more precise power to your computer than any
`other UPSyou can buy. Ournew UPScosts|essto install .
`and maintain
`- Want more? it fits in your offioe it's so quiet you‘ll hardly know its running And it's backed
`by a no-nonsense TWO YEAR WARRANTY from Elgar .
`.
`. the leader in quality power
`protection systems
`- Get the full story on our new l1Series UPS for mini computers.
`
`‘ion
`
`Call toll free: 800-854-2213, Dept. 8.
`
` ELGAR
`
`An Onan Operating Unit
`9250 Brown Deer Road
`San Diego. California 92121
`(619)450-0085 Telex 211063
`
`fBOOKLET OFFER
`r for what ever
`’ mana ernesds to
`our owerand
`_p@_Ii£. Call today
`i-€-
`;
` e Copy.
`
`
`
`
`Inn-on-my-an-nan-Iva-nuns-—-Inanncwvuunm
`
`
`
`51
`COMPUTERWORLD
`OCTOBER 2|. 1985
`.I_’
`
`IN DEP'|'I|/RELATIONAL DBMS
`
`values is largely or even
`completely factored out. For
`example, when there are 50
`distinct columns defined on
`U.S. currency, the data base
`is much easier to manage and
`manipulate if one avoids
`making 50 distinct declara-
`tions for U.S. currency.
`Before the relational disci-
`pline arrived, users had to
`make separate declarations,
`and as a result, many of the
`50 in the example would
`turn out to be incompatible
`with one another by acci-
`
`dent. The factoring of decla-
`ration that prevents these
`errors is achieved in Digital
`Equipment Corp.'s RDB,
`which has a concept of
`"global field definition.” But
`RDB fails to support domain
`constraints on certain opera- .
`tions, such as join.
`Another benefit of sup-
`porting the domain concept
`is that relational operators,
`such as joins and divides,
`that involve comparison of
`values between different col-
`umns can be constrained by
`
`the system. A DBMS can al-
`low data base values to be
`compared only when they
`come from the same domain
`and are therefore compara-
`ble from the semantic view-
`point.
`Such a constraint inhibits
`errors caused by interactive
`users of terminals who
`choose columns to be com-
`pared in such operations as
`joins. The wrong answers
`they obtain from these er-
`rors rarely uncover the er-
`rors themselves; meanwhile,
`
`unwise business decisions
`may be made based on these
`wrong answers.
`For various reasons, it is
`important to support as a
`qualifier in a command what
`I call “semantic override" —
`the ability to have the sys-
`tem ignore the usual compar-
`ison constraints. Users
`should be able to authorize
`this override qualifier sepa-
`rately from the operator in-
`volved and should authorize
`it rarely, reserving it chiefly
`for detective work.
`
`.
`.
`
`IO INSURE THAT EVERYONE
`CAN A(X3ESS DAFA, TRANSAMERICA
`USES I1\lTELLECT""Al‘ EVERY
`LEVEL OF THE PYRAMID.
`Transamericas lnformafion Center took
`over, developing custom INTELLECT
`applications for each department.
`Mr. Rahmqvist and Ms. Dah
`en
`trained a group ofTransamerica “veteran”
`INTELLECT users to go into the field to
`train the oompany’s 15 regional office per-
`sonnel in using the oompan s
`a ph-
`cations. 'llansamei-icas m¥E
`Support Staffhelps company personnel
`use INTELLECT more effectively _
`They’ve also devised several methods
`for enabling users to access their many
`INTELLECT applications. Easily under-
`stood menu screens help beginning
`users work faster, while experienced
`people can use an express mode for their
`application needs.
`“We’ve found that we’ve
`saved both time and money by
`using INTELLECT to obtain
`pertinent information.”
`—Ms. Sandra Dahlgren
`Anyone who can ask a uestion in
`everyday conversational English can get
`the information he or she needs. Imme-
`diately. You can imagine how much time
`a system like thatcan save an information-
`dependent organization like Transarnerica.
`Combine that with the ability to get more
`people into the system and you can see
`how INTELLECT has increased the
`pyramids power.
`Find out how INTELLECT can
`getyour organization into better shape.
`Write for a free demo diskette. Or for
`fast action, call AIC at (617) 890-8400.
`‘—---____--"I
`I I want to luiow more about
`I INTELLECT!
`| Name
`l Title
`
`relational data bases. All too
`frequently, I have seen situa-
`tions in which data base ad-
`ministrators with nonrela-
`tional DBMS failed to control
`their data bases adequately;
`consequently, they could not
`maintain a state of integrity.
`Domains
`
`Many users confuse the
`domain concept with the
`concept of attribute of a rela-
`tion or column of a table.
`Other people (often the ven-
`dors themselves) dismiss the
`domain concept as “academ-
`ic." My reply to them is: The
`atom bomb was also academ-
`ic!
`
`In fact, the domain con-
`cept is very important, prac-
`tical and simple. A domain
`consists of the whole set of
`legal values that can occur in
`a column. The column draws
`its values from the domain.
`Each column of a relational
`data base has precisely one
`domain, but any number of
`columns may share a do-
`main. There are several rea-
`sons why domains should be
`supported.
`For example, in a finan-
`cial data base, there may be
`as many as 50 distinct col-
`umns (possibly, but not nec-
`essarily, in distinct tables)
`defined around the U.S. cur-
`rency domain. Why repeat
`the definition of currency 50
`times? In data bases support-
`ed by nonrelational systems,
`I frequently observe many
`inconsistent declarations of
`value type for fields that
`were intended to have the
`same type.
`It is unreasonable to ex-
`pect a DBMS to store all the
`legal values in a domain, un-
`less there happen to be very
`few. However, it is entirely
`reasonable — and very
`worthwhile — to insist that
`a DBMS should store certain
`- values:
`I For each domain, a de-
`scription of the type of val-
`ues in that domain. This in-
`formation is global since it
`applies to the entire data-
`base, and it should of course
`be recorded in the catalog.
`I For each column, the
`name of the domain from
`which that column draws its
`values. This domain name is
`a reference to the global defi-
`nition.
`Of course, the domain de-
`scription can include range
`restrictions. For example, it
`could specify that quantities
`of parts in an inventory must
`not only be integers, they
`must also be non-negative.
`Furthermore, individual
`columns may include addi-
`tional range restrictions
`where these are semantically
`justifiable. In this example,
`the quantities of very expen-
`sive parts held in the inven-
`tory may be limited to some
`specified maximum.
`One of the benefits of sup-
`porting the domain concept
`is that, in cases where sever-
`al columns share a common
`‘domain, the declaration of
`the description of the legal
`
`At’li-ansamen'i:atheyseeINTELLECT
`as more than 21 stateof-th§-art natural
`language information retrieval system.
`They see it as a new way to do business:
`Giving all their end users——even those
`with no computer skills—instant access
`to more information than they’d ever
`had before.
`u
`iasi“I..‘I.‘ii‘§§§‘3.Y..‘°§‘.§‘Z§‘ia‘r’.§'§i°i‘;‘.2":t
`uaskouseiNTELLEcT
`b
`r
`1
`t
`1,,
`9931119 3 Ve1'Y POP“ 3' 00 -
`—Ms. Sandra Dahlgren,
`Information Center System
`Supervisor
`One Vice President uses INTELLECT
`to get the most current information on
`premiums and losses, and for longer term
`strategic planning with requests such as,
`“Give me the total June premiums and
`losses for each region.” Regional oflices
`use INTELLECT for a variety of tasks
`including asking INTELLECT to: “Tell
`me all about policy number 98579897.”
`Personnel keeps track of employee
`records, EEO compliance, human
`resource utilization, and more, by ques-
`tioning INTELLECT in plain Erighsh.
`Payroll, Claims, Underwriting, and Serv-
`ices also use a variety of INTELLFLIT
`applications. So from the top of the
`pyramid right down to the mailroom,
`Transamerica is using INTELLECT to
`work faster_and smarter.
`training,
`“After ]IlSt
`using
`people be
`INTEL ECT frequently
`“‘“~;,\ and effectively.”
`—Mi:Carl Rahmqvist
`—. 5?
`Aftersome initial
`
`implementation
`assistance from AIC,
`
`
`
`“We use INTELLECT because
`we want to give our users a
`better way to do business.”
`—Mr. Carl Rahmqvist, Senior
`Systems Manager, Information
`Systems’ fianwmefica
`.
`1'$“Ia'1°*’: Gf°"P
`G
`ransamenca nsurance
`roup is
`"=9°smdf°=thevmmld’ShaP¢d
`b”“”g¢“S”““‘TFm“m“““
`Corporation, and for beingorie ofthe
`oountrys leading commercial and per-
`sonal insurance companies. In some
`quarters, Transamerica is also recog-
`nized for its variety of innovative
`INTELLECT applications.
`
`-
`
`I C°'“”""’
`I Address
`
`I State Ep
`| relcpi-mei__.l.
`IOIS: VMICMS: MVST DBMS;
`Please send me information about
`INTELLECT.
`____ Please have-an Alcrepresentalive
`Contact me.
`cwum
`___________J
`ARTIFICIAL
`INTELLIGENCE CORR
`100 Fifth Avenue, Waltham, MA 02254 617-890-8400
`lNTEl.EI.'l"'u:ti1hriu'kdArIific’nllnIr!l'flaiceCiI'pc|I'n-i.
`
`IIIL
`
`
`
`
`
`COMPUTERWORLD
`
`OCTOBER 21. I985
`
`‘II DIPTII/RELATIONAI. DBMS
`
`'.. —
`
`_
`
`_
`
`.
`
`.
`
`
`
`
`
`52
`
`,
`
`Even when the domain concept is
`restricted to assigning types to data,
`it should not be confused with the
`hardware-supported data type. Con-
`sider the example of a data base
`listing suppliers, parts and projects.
`Suppose the hardware-supported
`data types of supplier serial numbers
`and D8112 Serial numbers are identi-»
`cal: each type consists of fixed-
`length strings of 12 characters. The
`system still needs to keep these two
`data types distinct and remember
`which columns are defined on one
`and which columns are defined on
`the other.
`If it can make these distinctions,
`then when a request comes in to
`delete or archive all records contain-
`mg X3 as a supplier serial number,
`the system can handle such a trans-
`action correctly. The system will not
`delete or archive any record that
`contains X3 as a part serial number
`and that also does not contain X3 as
`a supplier serial number.
`Today, such a data type is often
`called an application data type. The
`concept is supported in Pascal but in
`very few other languages that enjoy
`current use. The Pascal support does
`not, of course, include constraints on
`selects, unions, joins and divides.
`The domain concept is basically
`what makes all the meaningful se-
`lects, unions, joins and divides
`known to the DBMS. Thus, the do- -
`main makes the data base meaning-
`fully integrated, and it does so with-
`out prejudicing distributability.
`Contrast this with CODASYL links
`and IMS hierarchic links. They rep-
`resent the CODASYL and IMS con-
`
`,
`
`.
`.
`.
`0” ix.’
`-'
`
`_
`
`_
`
`.
`
`2
`g-“fig
`
`‘ -
`-
`
`should be a new CREATE DOMAIN
`command in SQL.
`.
`Fidehty
`Figure 1 shows fidelity to the 12
`rules by IBM's DB2 Cullinet Soft-
`ware, Inc.’s IDMS/R’ and Applied
`Data Research, Inc.'s Datacom/DB —
`examples chosen for their wide dif-
`ferences. These scores represent
`counts of compliance with each rule
`(score one for “yes" and zero for
`either “pa.rl:ial" or “no").
`Actually the information rule is
`so fundmnental to the relational ap-
`proach that a system's compliance
`with this rule should receive a much
`higher score than one. Weighting it
`as high as 10 would not be excessive,
`However, I shall avoid assigning dif-
`ferent points for different features,
`just as I avoided a fractional score
`for partial support of a feature: It is
`too easy to be subjective in these
`matters.
`DB2 scores quite well on the fidel-
`ity evaluation. Very few other DBMS
`score higher on the 12 rules, al-
`though some others score equally
`well. Both IDMS/R and Datacom/DB
`allow information to be represented
`in the order of records in storage and
`in repeating groups — directly vio-
`lating the information rule; In the
`case of IDMS/R, information may
`also be represented in links between
`record types (CODASYL calls them
`"owner-member sets") and also in
`"areas."
`Some vendors of nonrelational
`DBMS have quickly added a few rela-
`tional features —— in some, cases,
`very few features — in order to be
`
`mMs/R
`No
`N0
`N°
`No
`
`,
`
`“me
`; Information rule
`G“""a"'°°d *'’°°°55 “"9
`3 sV“°"‘“"° "°""“°"' °' ""''s
`4 A::LV:a°::,.,':,$based on resource
`
`082
`Yes
`Fania‘
`P‘’"'“'
`Yes
`
`Damwm/DB
`No
`"0
`N°
`No
`_ __
`2 elomnrehdglglsiée (1816 sublansuaee
`‘Q65
`n0
`so
`"'4'"
`" "" °
`°
`°
`°
`7 Hlgmevel mm‘ 33933! dame
`Yes
`N0
`"N:
`8 Physical data independence
`Yes
`Partial
`Partial
`lg hzdgrai d'antdae|ndedpendence
`PaNnia|
`No
`no
`.
`9
`‘V
`9°" W“
`°
`”°
`'3
`I‘ 5“ D'5mb""°" '.'.'.d°P.°"d"’»'-'5"?
`Yes
`N°
`N°
`12 Nonsubyefshn rule
`yes
`No
`"BIS
`score (I for yes, 0 otherwise)
`7
`0
`0
`
`Figure 1
`
`cept that a link “integrates an other-
`wise unintegrated data base," but
`they have several unfortunate re-
`strictions. Most importantly, they
`obstruct data base distribution be-
`cause of the constraints and com-
`plexity their data structures intro-
`duce into decisions regarding how
`the data should be deployed.
`A second serious drawback of
`links is that they are only paths.
`Generation of a result such as ajoin
`requires traversal of these paths by
`the application program. It seems su-
`perffuous to cite other difficulties
`with this concept.
`.
`Many relational DBMS and Ian-
`guages including SQL do not s_upport
`
`the concepts of primary key and for-
`eign key. I fail to see how these
`products can support the guaranteed
`access or the view updating rules
`without making the system aware of
`which column(s) constitute the pri-
`mary key of each base table.
`Furthermore, I fail to see how
`these products can support referen-
`tial integrity or the view updating
`rule without offering clear support
`for both primary keys and foreign
`keys. For example, in SQL, the GRE-
`ATE TABLE command should be ex-
`tended to permit the user to declare
`which column or columns constitute
`the primary key and which consti-
`tute foreign keys. In addition, there
`
`Need to recruit people for
`Your IBM systems?
`You'll find them reading the
`Computerworld Extra! on IBM.
`
`Published December 4th and closing October 25th,
`this special edition of Computerworld Extra! will take Cl
`hard look at IBM's products and strategies. Anyone
`working in the IBM arena will certainly review this issue.
`So if you're looking for pros in IBM systems, get your ad
`in this special issue and be surrounded by in—depth
`editorial on IBM.
`
`Computerworld Extra! will discuss IBM's strengths and
`weaknesses. We'll look at how SNA evolved, and how
`it will continue to evolve. And we'll discover whether
`IBM plans to provide 0 universal interconnect to SNA.
`Finally, we'll cover the alternatives, from PCs to main-
`fromes. LANs. Communications. And. of course, the
`compotobility issue.
`As you can see, this issue will have complete appeal
`for computer professionals working at IBM installed
`sites.
`
`To reserve space coll Al DeMille, Notional Recruitment
`Sales Manager. at (800) 343-6474, or (617) 879-0700
`in Massachusetts.
`There is no special classified section: all recruitment ads are
`considered display advertising for this issue.
`
`
`T S 0 users
`DSM'"(Data Set Manager)
`improves TSO Performance
`.......__ ,,,._.__.._...... ...................
`
`T
`
`.q—.-.-..—'-.,—'—----—--‘--.-.------—---.-0... 5.. Mung.’ ...............................
`PFK ) I-Y} 2-BOTTUA. 3-PF.
`IIPB. 5-FDND. I-QUIT. 7-FHINV. I-PFK. I-I-ELF
`CATALOG PllEFI)(- JONES
`(FF) alas-I cnmm-nd Iruu: Arn-
`<~-DSHANE-->vDL5EII ALLDC UEED Ex ORG IIECFM BLKSIZE LRECL LASVREF
`06/13!!!
`FB
`Jun
`an
`AVLDNIL
`TSDOID
`3
`1
`SPF
`F!
`3|?!)
`an
`ufliialau
`APRDJJEXT
`E
`VEDDIZ
`no/lama
`FB
`3|2D
`B0
`150i.'lun
`BIICDBOL
`F§
`DB/23/3'4
`VB
`3I2|'.|
`235
`Imam
`XL CFZIJEXY
`Wlzulau
`Fl!
`IODII
`GD
`150005
`REN DEN.FDRV
`iu/Dz/nu
`FE
`sun
`on
`cosv Ez.AsM
`vsoonz
`CI-u I you
`Fl!
`:ll2I'.I
`nu
`YSDDOG
`sue
`I=Mu)I.r:NrI_
`vs
`:n2n
`clllliusu
`PBNY G2l.l.Ll5T
`T500173
`VB
`:4 I 20
`235
`lull wen
`conv HOLJEX1
`YSDI708
`u
`D
`an/oz/an
`ISJDIO
`ruse IMoDEL.LDAI:I
`150002
`LM
`.x:I.I.r:~1L
`FQ
`3| 20
`B0
`09/ I5!!!
`FB
`3| :0
`Du/2013-I
`an
`FREE I<A.I:cieoI.
`Iaanaiu
`VB
`aim
`ID/2:/an
`VSODGS
`----
`1m.PI.I
`
`
`III
`
`I
`I
`I
`I
`Cl
`I
`I
`I
`I
`
`lHl33'BJiJlEJlJlJHHi
`
`O DSM provides full-screen displays of cataloged files and supports TSO
`functions with simple. cast-to-remernber command names.
`0 DSM support: SPF, FSE. CLlSTs and Command Processors. DSM's
`lludble design allows users to define their own commands.
`
`O DSM file commands include: DELETE, RENAME, COPY. SUBMIT, PRINT.
`COMPRESS, Release unused space. List PDS members and deallocate.
`
`0 DSM supports PDS member displays. HSM and Mss.
`
`0 Security? - Our user unit lets you control all functions.
`Call or Write for details:
`
`Applied Software. Inc.
`340 u.s. Highway *1, Suite 250.
`p.o. Box 13027. North Pam Beach, FL 33403
`(305) 6264818
`
`
`
`THE WORKSTATION
`THAT PIGGYBACKS
`0 THE PC.
`, -
`_
`Everyoneknows thePC isapowerfultool.What
`l-<3
`\'I';),§,/} everyonedoesn’tknow is thatmany office
`
`((
`
`«
`
`
`\
`
`’ ~»
`‘
`
`
`
`workers who use personal computers spend
`less than an hour a day actually working on them Which
`makes it next to impossible to cost-justify an individual
`U PC.for everyone in your company.
`’
`That's where Davox comes in Davox takes the
`power of the IBM PC and allows that power to be
`shared by multiple users. With Davox worksta-
`tions, a number of users (typically 4 or 5)
`. spread throughout a network can access and
`I utilize the full computing capabilities of a
`.' single PC. Aft afraction ofittl)1e cost1t%y(%ur
`company 0 going ou an
`s or
`everyone.
`_
`_
`i But that’s just the first reason for going
`‘v
`with Davox. Davox integrated voice!data work-
`stations let you access your IBM mainframe, DEC
`mini’s, dial-up databases or IBM PC’s. And
`Davox’s 16 “Smart Buttons” let you access
`these different computer resources with a
`single keystroke. You can even dial a phone
`number directly fiom the screen with a
`single keystroke And DavoxNet’s twisted-
`‘
`pair telephone wire networking design
`J] eliminates expensive coaxial wiring.
`To find out more, call Mike Giltner
`at (617) 667-4455 or toll-free at
`1-800-343-1152. Or mail in the
`coupon below.
`
`__
`
`~)
`
`-
`
`=
`
`,
`
`V
`
`uying
`
`,
`fig‘.
`e’
`
`3!
`
`
`
`
`
` ‘.5 /
`
`
`
`DAVOX"
`
`We are data and voice systems.
`We do it all. And we do it now.
`‘ll
`F: Yes! I’d like to find out how to
`piggyback on a PC.
`
`Company
`
`I I
`
`]
`
`I
`
`“*7
`.
`--=
`I-
`
`
`I
`I
`I
`Send to: Davox, 4 Rederal Street,
`illerica, MA O182L
`..,m,c.,-1
`
`
`
`
`‘
`=
`//-A/ 7
`“/////3 _
`..
`..
`.
`
`.
`v
`‘/-.5 /J J A!
`9
`I .?§'//,
`57/;
`‘/1. 17». //n /K’;/z.“(7z.'(
`.
`
`
`.. ggi-,3; /_,'_’¢_,,i::;i.k,.o;1.,',,*.4,
`i!,9;!,!> as
`z ,
`T
`’
`¢
`/‘
`r
`
`47;! .7l'IQ'I%/‘I/J >//4317//Al’//.
`
` !'Wa."WflI'
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`NOWWE CAN SELL vou THE WHOLE LINE '
`or EQUIPMENT YOU NEED.
`
`____.._————---"""“"""‘_""“
`
`INCLUDING THE LINE.
`
`—I
`I-For more information, contact
`NYNEX Business information Systems
`I
`Po. Box 655, White Plains. NY 10604-0655.
`I Or call:
`I-800-346-9X9X ... |
`call me.
`El Pleasehave a NYNEX Account Executive
`I
`
`I I
`
`:..?..:"—,————-———-—
`
`
` I
`
`
`c.,
`
`amaze:
`_ __: _l
`
`go beyond those of any single manufac-
`turer. Because we’ve selected all the
`best products from all the best companies.
`in fact, we offer a full range of networking.
`voice and data commu nications, and
`computer products that work together in
`one neatly integrated system.
`The result? An efficient system that can
`meet all your needs today and still grow
`tomorrow. A system designed, installed
`and maintained by a team of professionals.
`And backed by a company with over 100
`years of experience that will continue to
`lead the industry far into the future.
`Togetourteam workingon all youroommuni-
`cation needs, all ittakes is one phone call.
`Then compare us with the competition.
`You'll find we're ahead by a wide margin.
`Not just a fine line.
`
`The shortest distance between you
`and the best business communications
`system is to head straight to us. NYNEX
`Business Information Systems.
`Because we have everything it takes to
`build the best customized communica-
`tions system for your business. The ex-
`perience, the technical know—how, the
`products and the people.
`As authorized agents of New York
`Telephone and New England Telephone,
`we‘ll provide you with a network that deliv-
`ers the most reliable. practical and cost-
`etfective voice and data transmission
`services available today. We're also in the
`perfect position to build your company a
`custom network utilizing the latest fiber
`optic and microwave technologies.
`What's more, our equipment solutions
`
`BUILDING BUSINESS COMMUNICATIONS FROM STRENGTH.“
`
`NYNE..‘-'
`Business Information Systems
`@
`“Service Mark ofNYNEX Corporation. @1985 NYNEX Busines iriforrrlmion Systems.
`
`
`
`60
`
`COMPUTERWORLD
`
`OCTOBER 21, 1985
`
`
`IN DEPTH/RELATIONAL DBMS .
`
`
`Evaluation against the relational model
`(by DBMS)
`
`Relations
`Base tables
`View tables
`Query tables
`Snapshot tables
`Attributes
`Domains
`Primary keys
`Foreign keys
`Theta select
`Project
`Theta join
`Outer theta Join
`Union
`Outer union
`Intersection
`Division
`Set difference
`Relational asslgkient
`Maybe there select
`Maybe theta join
`Maybe outer theta join
`Maybe divide
`5/0 theta select
`5/0 theta join
`5/0 outer theta Join
`
`DB2
`Yes
`Yes
`Yes
`Yes
`No
`Yes
`No
`Partial
`No
`Yes
`Yes
`Yes
`No
`Yes
`No
`Yes
`Yes
`Yes
`Yes
`Yes
`Yes
`No
`Yes
`No
`No
`No
`
`IDMS,/R
`No
`Yes
`No
`No
`No
`Yes
`No
`No
`No
`Yes
`Yes
`Partial
`No
`No
`No
`No
`No
`No
`No
`No
`No
`No
`No
`No .
`No
`No
`
`L:/0 divide
`Entity Integrity
`Reieremiai iutesfitv
`User-defined integrity
`Total score against relational model
`(.1 for full support. 0 otherwise)
`
`No
`No
`Partial
`No
`No
`mi-rial
`
`Partial
`No
`15
`4
`
`Datacom/DB
`No
`Yes
`No
`Yes
`No
`Yes
`No
`No
`No
`Yes
`Yes
`No
`No
`No
`No
`No
`o
`:0
`No
`No
`No
`No
`No
`No
`No
`NoN
`o
`No’
`No
`‘No _
`5
`
`-
`
`_
`
`ibtaiscoreon 12 rules
`Gran! mi score
`9‘ fidelity
`
`1
`23
`4-6
`
`_0‘”
`4
`B
`
`H
`
`0
`5
`10
`
`You can reach over
`50,000 computer
`professionals in
`Norway.
`
`Advertise in CW Communications’ Norwegian publications and sell
`your products directly to Norway’s thriving computer market. According
`to international Data Corporation (l'DC), the world's leading information
`industry research firm, Norway's total DP expenditures for 1983 were
`$1.6
`IDC forecasts that DP expenditure will reach $3 billion by
`1989. The
`computer market is also a luaative one with
`expenditures in 1983 of $58
`CW Communications has two publications covering the Norwegian
`market; Compuierworld Norge and PC Mikrodatu.
`Twice a month, 20,000 MIS/DP executives read Computer-world Norse.
`Modeled after Cornputerworld, its sister publication in the United States,
`Cnmpuierworid Norge is written for the professional end-user and reports
`on the most timely news concerning new products, new software
`applications and market tn.-nds and opportunities.
`And every month, 30,000 professional and home PC users read PC
`Mikmdata. The editorial focuses on a wide range of topics including the
`latest information on the PC market, new software, hardware, inter-
`views, buyers guides and instructions for the PC.
`CW International Marketing Services makes advertising your products
`in Norway, and around the world, easy. We have over 50 publications in
`more than 25 countries. For more infomiaiion on our wide range of
`services, call Diana la Muraglia, General Manager, CW Intemalional
`Marketing Services, at (800) 343-6474. On the West Coast, call Isabella
`Barbagallo at (415) 328-4602 or (800) 277-8365. Or fill out the coupon
`below and mail today.
`
`W COUM|.|HI(ATflUL K.
`
`Please send me more information on:
`El Compuierworid Norge
`[II PC Mikrcdata
`El Your other fordgn publications
`NIH!
`Diana [.1 Muraglia
`Tulle
`Gmenrl Manager
`CW international Marketing Services camp“
`375 Cochituate Road, Box sea
`Framingham. MA 01701
`
`’
`
`(as implemented) already possesses
`is underemphasized. The entire
`range of SQL implementations from
`the large mainframes down to the
`micro is not adequately addressed.
`Finally, ANSI ought to extend
`presently supported SQL to a version
`that fully supports the relational
`model, including distributed data
`base. At the very least, ANSI should
`generate a statement- of direction ad-
`equate to permit vendors to extend
`the fidelity of their products without
`risking incompatibility with some fu-
`ture standard.
`'
`Extensions of SQL that provide
`this support now can be forecast in
`detail and with some reliability. Any
`standard adopted now should not
`makesthese extensions impossible or
`even difficult in the future.
`
`Three buying factors
`Any buyer confronted with the
`decision of which DBMS to acquire
`should weigh three factors heavily.
`The first factor is the buyer’s perfor-
`mance requirements, often expressed
`in terms of the number of transac-
`tions that must be executed per sec-
`ond. The average complexity of each
`transaction is also an important con-
`sideration. Only if the performance
`requirements are extremely severe
`should buyers rule out present rela-
`tional DBMS products on this basis.
`Even then buyers should design per-
`formance tests of their own, rather
`than rely on vendor-designed tests or
`vendor-declared strategies.
`The second factor is reduced costs
`for developing new data bases and
`new application programs. Relation-
`al DBMS provide significant reduc-
`tion in these costs, when compared
`with either the CODASYL or hierar-
`chic approaches. Fourth-generation
`languages are no substitute, al-
`though they may provide some addi-
`tional productivity.
`The third factor is protecting fu-
`ture investments in application pro-
`grams by acquiring a DBMS with a
`solid theoretical foundation and reli-
`able support for high productivity
`and distributability. In every case, a
`relational DBMS wins on factors two
`and three. In many cases, it can win
`on factor one also — in spite of all
`the myths about performance.
`Then the question arises: Which
`relational DBMS? The system chosen
`should not only be a DBMS with a
`good percentage of fidelity to the
`
`At the very least,
`ANSI should
`generate a
`statement of
`direction to permit
`vendors to extend
`their products
`without risking
`incompatibility
`with somefuture
`standard.
`
`
`relational model, but should be ex-
`tensible at some future time. Ideally
`a good DBMS will be extended soon
`to provide 100% support without
`logically impairing the customer’s in-
`vestment in application programs.
`Buyers should be cautious with
`vendors that make