tesdata_acm_72
TRANSCRIPT
8/3/2019 tesdata_acm_72
http://slidepdf.com/reader/full/tesdataacm72 1/8
8/3/2019 tesdata_acm_72
http://slidepdf.com/reader/full/tesdataacm72 2/8
INTRODUCING
The rapid growth of computer usage andesulting expenses to enterprises and
government has created a requirement to
pursue means for effectively managingcomputer hardware and personnel assets,plus to obtain optimum usage of each.
The initial attention to computer costscreated the "third party" computereasing in the early 60's. This was
ollowed by the "plug-to-plug" compati-ble perhipheral industry. The most
recent development of advanced meansor measuring, simulating, optimizing
and scheduling of computer hardwareand programs is rapidly gaining accept-
ance by the influential users of com-puter systems, and is the latest move
oward obtaining optimum performance.t is these latter areas that are the focus
of Tesdata Systems Corporation.
Tesdata is a leader in this field of
providing products and services for
improving computer performance. The
broad line of its products makes the
company unique in this highly special-
ized segment of the computer field, and
the success of Tesdata demonstrates the
rapidly growing acceptance of its capa-
bilities and products among major com-
puter users throughout the world.
The company's wealth of practicalexperience in advanced data manage-ment and communications systems cou-
pled with its expertise in computermeasurement, simulation, and scheduling
provides the diversity of skills essentialin providing practical assistance to some
of the most advanced users of computersystems. Tesdata clients include leading
corporations among airlines, utilities,banks, insurance companies, railroads,
state and federal governments, consult-ants, and numerous types of manu-
facturers.
The operational structure of Tesdata
consists of five basic areas, with each
group appropriately staffed for handling
its current base of clients, new custom-
ers and prospects, plus product improve-
ments. The five areas and their capabili-
ties are:
ANALYTICAL SERVICES
This division of the company is heavily
experienced in the area of systemarchitecture and programming of on-line,
real time data management systems. Itsmajor assignment during the past year
consisted of an average of ten men indirect support of implementation ofcapabilities for management of data in a
real time, on-line environment utilizingHoneywell G-635 Systems at a federal
government installation. A second effortwas to perform system architectural
design in the area of data communica-tions and data management systems for
an IBM 370/155 System in a largecommercial account.
MEASUREMENT SYSTEMS
This department is responsible for themanufacturing, installation, support and
maintenance of the X-RAY ComputerPerformance Measurement System and
for contract measurement services forusers throughout the world. X-RAY
Systems and services have been providedto customers using IBM, UNIVAC,Control Data, Honeywell, RCA and ICL
computer systems. Department person-nel have broad systems and applications
backgrounds to complement indepthmeasurement analysis experience.
OPTIMIZATION SYSTEMS
This department is responsible for prod-
ucts, services and instruction in COBOLprogram optimization. Department per-
sonnel install, support, and maintain theSTAGE II COBOL Source Program
Optimizer, conduct seminars in COBOL
optimization techniques, and offer
proven by-mail service for COB
source code optimization to meet ctomer one-time or occasional requ
ments.
SCHEDULING SYSTEMS
This department is responsible for installation, maintenance and appli
tions support of the corporation's schuling products. DEADLINE is unique computer center scheduling s
tem currently installed in data procsing operations of major corporati
throughout the country. A second pruct, STREAMLINE, is a compu
system job stream scheduler for increing throughput performance and
turn-around response.
SIMULATION SYSTEMS
This department is responsible for installation, maintenance and suppor
the CASE Simulation System and performing contract simulation servic
As the corporation's first major prodCASE has been accepted by over hundred customers in its three-y
history. CASE IV, the latest verstogether with the recently develo
SMF/CASE Input Processor for automing workload definition, has convince
number of experienced simulation usto recognize the system as being
most advanced available from indutoday. The experienced staff supp
CASE installat ion and service actiand performs a continuing functionproduct improvement and library
hancement.
8/3/2019 tesdata_acm_72
http://slidepdf.com/reader/full/tesdataacm72 3/8
Any decision to increase system effi-
ciency is as meaningful as the dataavailable on current system performance.
Similarly, the decision will have valueonly if its success or fai lure can bemeasured once it has been implemented.
Making a performance decision based onfragmentary measurement data on cur-
rent efficiency establishes a low proba-bility of success in improving operating
performance. Of course, given an inade-quate measurement capability, it is
equally improbable that efficiency in-creases (or decreases, for that matter)
will be accurately detected after aperformance decision has been executed.
X-RAY (Execution—Recorder Analyzer)removes the guesswork involved in
making constructive system decisionsand verifying their effectiveness. X-RAY reports actual operating performance on
which to base sound system improve- ment decisions. The validity of these
decisions, once implemented, can thenbe shown using X-RAY to measure their
precise effectiveness. Making wise per-formance decisions is a prerequisite to
maximizing system efficiency. There isno better way to ensure success in
making these decisions than to applycomprehensive measurement using the
most capable system available: X-RAY.
WHAT WILL X-RAY MEASURE?
Virtually everything. X-RAY measures
all areas of computer system activity:• Equipment• System Software
• Appl ication Programs
• Data
Unlike other measurement systems,
X-RAY is not restricted to monitoring just your hardware. Of course, X-RAY
does report on equipment utilizationhighlighting device usage and imbalances
within your configuration. But X-RAYwill also measure operating system
activity providing demand and overhead
statistics and isolating areas of highactivity. Applications programs may alsobe monitored for structure and codeefficiency, Data handling can be exam-
ined in depth to report on accessoverhead, file organization and device
contention.
WHAT RESULTS ARE POSSIBLE?
X-RAY pinpoints multiple areas ofoperating inefficiency so that many
system improvement measures such asthe following may rapidly be applied:
• Configuration Balancing
• Operating System Real location
• Problem Program Restructur• Data Base Reorganization
• Program Code Optimizat ion
X-RAY gets results. In a single moni
ing session at a Government agency
was discovered that 40% of the ti
spent by a major, processor-bouapplications program was spent in o
10 instructions of the program—
excellent demonstration of isolation
high activity program code. Sim
recoding did away with this signific
bottleneck.
At a commercial installation, X-Rpinpointed an inefficient direct
search method causing overall throuput degradation exceeding 20%
incredibly minor change was effecteda few minutes time remedying
situation.
8/3/2019 tesdata_acm_72
http://slidepdf.com/reader/full/tesdataacm72 4/8
STAGE IICOBOL SourceProgram
Optimizer
Until recently, users have been paying
for the relative inefficiency of the
COBOL language, resorting to hardware
upgrades and additional operations shifts
rather than attacking the problem at the
source: the SOURCE program.
The gap between technical programmingstaffs and non-technical management has
resulted in the frequent supposition, by
COBOL users, that there is simply
nothing that can be done to reduce the
unnecessary overheads of their language.
For the first time since COBOL'sadoption, however, the language has
come under critical scrutiny, and simplemethods of overhead avoidance have
been developed. The results in decreased
run times, core reductions and improvedproductivity have been dramatic.
WHAT IS STAGE II?
STAGE II is a COBOL program. It isthe automated implementation of tested
optimization methods, analyzing sourceCOBOL programs, producing a diagnos-
tic listing and, where possible, generating
hardcopy correction cards. Stage IIinputs the target COBOL source pro-
gram and produces a card image correc-tion file for reinsertion into the original
source program or library. These correc-tions are made for every inefficiency
which can be corrected automatically in
the working-storage section. A diagnosticlisting is produced for each correction
informing the user of the deficiency atthe point where it occurs.
STAGE II produces an efficiency log
which indicates additional modifications
required for effecting total optimization
of the program.
STAGE II includes a highly efficient
alpha cross referencing sub-system whichminimizes programmer interface and
adds to program documentation.
STATE II allows for the insertion ofinstallation parameters and will diagnose
the use of any COBOL features whichare pre-selected as being in violation of
an installation's standards of program-
ming.
HOW MUCH TIME DOES IT TAKE TOOPTIMIZE A PROGRAM?
STAGE II does the complete analysis
and correction card output in one-third
compile time. Your programmer inserts
the new source cards, looks over
STAGE II's suggestions for total optimi-
zation and decides which of these to
implement. He then makes the procedu-
ral coding revisions—this takes from ten
minutes to an average of two to four
hours, depending on the number of
source cards—and the program is ready
to compile, test and run.
WHAT ABOUT TRAINING?
The lease of STAGE II includes an
intensive one-day course in techniquesof source program optimization for
COBOL. Up to fifteen participants are
trained, at your installation, by anexpert in throughput optimization. Your
key personnel learn how to write moreefficient, faster running COBOL pro-
grams. With the aid of STAGE IIoutput, each participant optimizes pro-
duction programs at the workshop
session. Text and materials are providedfor each participant and emphasis is on
actual use of the STAGE II package. Inaddition, your programmers become
optimizing programmers, fully trained in
the following: methods of more efficientdata definition and coding techniques,
avoidance of improper proceduralcoding, establishing "best type" and
break-even points in all efficiency meth-
ods, defining data patterns and methodsof using them to best advantage, deriv-
ing simple formulae for computing I/Orequirements, knowing where and when
to optimize, maximizing core conserva-
tion, establishing proper file organiza
and block sizes, and determining ac
COBOL verb timings.
WHAT RESULTS DOES STAGE II G
STAGE II has demonstrated that a
70% of all COBOL programs are hi
optimizable. STAGE II maintains
average of 25% run time reduction,
users are reporting reduction of up
47%.
STAGE II also accounts for size
decreases in core. Savings in one gram alone can justify the total inv
ment.
STAGE II serves as a continuing ed
tional tool. Quite separately from skills gained in the initial training,
software becomes a guide to beginprogrammers in demonstrating pr
coding techniques and introdu
simple corrective measures to be immented in their own programs.
STAGE II functions as an autom
"standards manual." The software
sures staff-wide observance of effi
coding methods, previously ignored
COBOL programmers.
STAGE II enables management, fofirst time, to determine the leve
inefficiency in its inhouse progrthus bridging the gap between techand non-technical staff. It also prov
the programmer with the necesinformation to correct the problem
its source-the SOURCE program.
8/3/2019 tesdata_acm_72
http://slidepdf.com/reader/full/tesdataacm72 5/8
DEADLINE IIComputer CenterScheduling System
Too frequently the daily problems of
computer operations tend to obscure the
basic requirement of returning com-pleted work to the user at the desired
time. The processing of a job from thetime it is received until it is returned
involves a significant number of activ-
ities in addition to the primary functionof computing. The most powerful com-
puter available cannot be used effec-tively if input is not supplied in a
proper and timely fashion, and output isnot delivered to the user when it is
required.
This last point is the most practical
gauge of how well the center is
performing its job. In many organiza-
tions the questions of "where is my
work, when can I get it, or why can't I
get it now" are most familiar. In
attempting to answer these questions the
data processing manager is faced with a
most frustrating task because generallyhe doesn't know. Indeed the attempt to
find a solution for the users leads to
other questions from his management
such as "why do you need more people,
more equipment, more overtime?"
There is a mechanism available to you
that answers these questions, and many
more, before they are asked. TheDEADLINE Computer Center Schedul-
ing System tells you what jobs will becompleted by your desired completion
time, which will not, why not, and
allows you to make corrections beforethe fact!
DEADLINE is a management control
system for the monitoring and adminis-
tration of the work flow and resources
throughout a computer center. It pro-
duces predictive schedules and reports of
how the work should be distributed
across resources to meet pre-established
completion times.
In addition to functional scheduling, the
system instills a discipline throughoutthe center. All activities are coordinated
and documented from the time workleaves the originator until it is returned.
Summary reports show how personnel
and equipment are employed, the totalworkload for the center, the total
resource capacity, and whether or not
work will be completed when desired.
JOB PRIORITIES
DEADLINE incorporates a unique, yet
simple, hierarchy scheme for job priori-
ties. Priorities may be user specified or
computer generated. They may be based
on minimum throughput time, on meet-
ing scheduled deadlines, or on accept-
able waiting periods for low priority,
nonproduction jobs. The DEADLINE
priority control system automatically
reviews and updates priorities in accord-
ance with user-specified objectives.
A COMPREHENSIVE SCHEDULER
DEADLINE produces complete sched-
ules for all computer center jobs andactivities. Job schedules trace work
through each activity area (keypunch,
EAM, compute, print, report control,etc.) while activity schedules provide
shift schedules for operating and super-visory personnel. Schedules present start
and completion time for each job task,
manpower/machine assignments, and asummary profile of resource utilization
against resource availability.
Flexible design provides for selective
breakouts of schedules most meaningful
to specific organizational and activity
groupings. Schedules for specific
machines are also produced where re-
quired.
A PLANNING TOOL
As a planning tool, DEADLINE
present hardware capability agai
data center's unique workload in that are most meaningful to ma
ment, namely, the extent to hardware capability directly affect
completion of jobs to schedule, thr
put time, machine/manpower utiliand leveling, and overtime and
requirements. DEADLINE providebasis for continual examinatio
equipment capabilities in termchanging user needs. DEADLINE
vides data useful for equipment e
tion studies and for personnel sdecisions, and in determining exp
capability.
MULTI-PROGRAMMING PROCES
DEADLINE incorporates all the fe
to consider constraints and capaassociated with multiprogrammingcessing with new third generationputers. Both overall program proccapability and dynamic allocaticapability are considered with amum of data inputs.
8/3/2019 tesdata_acm_72
http://slidepdf.com/reader/full/tesdataacm72 6/8
STREAMLINEComputerOperationsScheduler
The primary goal of computer opera-tions is to provide timely and responsivereturn of work submitted for processingwith a minimum investment in dataprocessing equipment, software develop-ment, operating procedures, and stafforganization. The attainment of this goalhas been made substantially more diffi-cult with the highly complex configura-tions available with current data proc-essing hardware and software.
The rapid enlargement of data proc-essing capabilities and flexibilities hascreated unique problems for data proc-essing management. Because of thecomplexity of current data processingsystems and the fact that millions ofdollars and hundreds of people areinvolved and affected, no longer is acorrect data processing decision readilyapparent. Paradoxically, the personnelsupplying the information needs of thecorporation are obtaining the least infor-mation from the new capabilities andresources. It is an undeniable fact today,in the typical data processing environ-
ment, that a serious lack of managementtools exist to plan, schedule, and controlproduction processes in a data proc-essing center. Without these tools andgiven this environment, rational dataprocessing decisions cannot be made normaximum operating efficiency achieved.
To be responsive to these needs, a dataprocessing planning, scheduling, and con-trol system must be closely tailored tothe unique environment in which it is tooperate. STREAMLINE is just such asystem. Developed and tested in acomputer center environment, STREAM-
LINE provides planning and controlcapabilities the computer center managerand his staff now need.
The implementation of STREAMLINE isresponsive to the areas of managementconcern:
• Reduction of job reruns• Reduction of late reports• Improvement in total center
thruput• More cost effective computing• Evaluation of changes in
resource or workload networks
• Control of tape/disk proliferation
STREAMLINE provides for the timelymanagement of a computer systemworkload. The STREAMLINE Systemprovides automated scheduling of themajor elements of the center environ-ment including processors, peripheralunits, and data sets. The schedulerproduces a series of predictive reportsdesigned for al l levels of data processingmanagement. These reports establish acommon vehicle of communication forcontrolling the workflow. The utilitiesprovide reports on the inventory andutilization of the center.
The capability for detailed workloadspecification and control techniquesenables significant cost savings by im-proving effective utilization and in-creasing throughput on the most costlyresources within the computer center.The system is designed to achieveplanning, scheduling, and performanceimprovement objectives within the con-text of a computer center.
A PLANNING TOOLAs a planning tool, STREAMLINEpredicts future performance, permittingmanagement to review in advance theperformance within the center. Notifica-tion of peaks and valleys in theupcoming workload allow for earlyrecognition and subsequent correctionsof imbalances. STREAMLINE has thecapability of specifying a computercenter network and will support evalua-tion runs to determine the effect ofequipment variation on an operationallydefined workload.
A SCHEDULING TOOLAs a scheduling tool, STREAMLINEprovides multi-level management reportsfor an installation's defined schedulingperiod. These reports are then dis-
seminated to cognizant staff personnelfor establishment of common objectivesand specific responsibilities. Thescheduling interval is controlled by asimple input parameter and can varyfrom one hour to one week. All work tobe scheduled is structured into jobs andthen tasks within jobs. The jobs providethe majority of the scheduling param-eters. The tasks complete this data andadd the performance data used forintra-processor scheduling (multi-programming). Schedules are producedfor individual processors, groups of proc-essors, or external users, depending onthe needs of management.
PERFORMANCE IMPROVEMENT CAABILITYAs a performance improvement capbility, STREAMLINE provides for tscheduling of compatible tasks aleveling of workflow across all proessors within the center. The systmaintains a resource map and considtask performance data and resource quirements on a processor prior scheduling. This allows jobs which crun on several processors to scheduled on the one which can meffectively run the job. For computer operator, the system provid
a scenario of his shift workload, speccation of his configuration, availabiof resources for unscheduled work, athe initiator configuration to process work. By knowing in advance specific configuration and time phasworkload, the operator has consideramore information on which to base operational decisions.
Through the structures established STREAMLINE for controlling, plannand scheduling of the total center woload, other valuable management dataderived. This data consists of:
• an inventory report of quantity and attributes tapes, disks and other simresources being used in center.
• a hardware inventory of schedulable resources, specing in addition their connecity, status, and use.
• a common data library of scheduling, performance adescriptive information for
jobs/tasks being scheduled.
These reports when complemented wschedule reports are of considera
value to data processing managementmany critical decisions.
8/3/2019 tesdata_acm_72
http://slidepdf.com/reader/full/tesdataacm72 7/8
CASESimulationSystem
Data processing decisions are becomingmore difficult to make as computersystems and applications grow in com-plexity and sophistication. At the sametime, these decisions can affect pro-
cessing costs more significantly than everbefore. Failure to recognize changingresource utilization or to anticipate theeffects of new computing requirementscan impact system performance and
responsiveness with corresponding in-creases in operating costs. The lack ofintelligent system management and plan-ning can create a data processingnightmare.
How can more effective decisions bemade in the complex environments ofmultiprogramming, multiprocessing,
real-time, time-sharing, remote job entry,and total information systems? How are
supporting facts and figures to beobtained for the evaluation and selectionof a new computer system? What are
the performance consequences of install-ing a new remote terminal system? What
bottlenecks can be expected? How isresponse time affected? How will an-other manufacturer's peripheral equip-ment perform on the existing system?What is the optimum configuration to
support a new application?
Clearly, to make wise decisions theproper tools should be used rather thanbasing decisions on partial data, rough
approximations or sheer intuition.
CASE (Computer-Aided System Evalua-tion) is the tool used at many data
processing installations to assist in mak-
ing complex system decisions. CASE
streamlines the decision-making process
by providing a capability for accurate
simulation of present and proposed
systems and applications.
Important to CASE users is the abilityto simulate data management systemssuch as IMS2 with its dynamic buffering
characteristic. CASE clearly indicates theprobability that a requested recordexists in the buffer. It includes theprocessor overhead for the data manage-ment system.
FEASIBILITY STUDIES
A CASE simulation quickly and accu-
rately establishes the feasibility of a
basic design approach. The study might
involve either a new design or the
incorporation of new features into an
existing system or application.
SYSTEM DESIGN
Through the use of CASE a largenumber of system or application desiqnalternatives can be evaluated with mini-mal manpower and time investment.New designs can be explored with easeto improve existing systems and applica-tions.
COMPUTER SYSTEM SELECTION
CASE provides the abil ity to perform
detailed competitive analyses of com-
puter systems proposed by various
manufacturers relative to each other and
in response to fixed processing specifica-
tions. Optimum configurations for each
manufacturer may be determined to
meet existing or projected workloads.
CONFIGURATION MANAGEMENT
For on-going configuration management,control and planning, CASE may be
used to ascertain performance interrela-tionships of new equipment, system
software and applications. Using CApresent and projected workloads maspecified in machine independent
which facilitates rapid, automatic, tem redesign for analysis of vaconfigurations.
CASE/SMF INPUT PROCESSOR
To ease the input definition to CA
an auxiliary package is offered wautomatically generates CASE from SMF accounting tapes. The Cinput is extremely accurate, ass
precise simulations. But most imtantly, minimum time and effort required to prepare simulations.
The data created by the CASE/Input Processor may be used to sim
any hardware configuration desired. CASE input data is completely macindependent.
NETWORK ANALYSIS
CASE provides a thorough teleprocenetwork analysis unrestricted by number of communication compon
defined in the system. A thoroanalysis provides network compoutilization and queuing information.
8/3/2019 tesdata_acm_72
http://slidepdf.com/reader/full/tesdataacm72 8/8
TESDATAPRODUCTSUSERSINCLUDE
Amoco Production CompanyAmpex Corporation
Blue Cross AssociationCelanese Corporation
Control Data CorporationCIMA InsuranceVirginia Division of Motor Vehicles
Ernst & Ernst
Honeywell Information SystemsITALSIEL
Northwest Computer ServicesRalston Purina
Drugs & Chemicals, Inc.
U.S. Forest Service
Ontario Hydro
Western Electric
Naval Command Systems Support Activity
(NAVCOSSACT)
U.S. Army Computer Systems Command(USACSC)
American Telephone & TelegraphIllinois Bell TelephoneBoeing Computer Services, Inc.Naval Ordnance Systems CommandBankers Security Life InsuranceServi-Data ABPANDATA NVScholastic Magazines, Inc.Sandia Corporation
U.S. Naval Construction Batallion CenterARINC Research CorporationTransamerica/ISD
Firemans Fund Insurance Co.The Cleveland Electric Illuminating
CompanyConsolidated Edison Co. of New York, Inc.Del Monte CorporationState of OklahomaFraser EspanolHoneywell-ltalia
OlivettiDetroit Edison
Grumman Data Systems
Hallmark Cards, Inc.
Lockheed-Georgia CompanyMcGraw Hill, Inc.Northern States Power Company
Pacific Gas and Electric Co.Southern New England Telephone CompanySpring Mills, Inc.
Transamerica Corporation
Trans-World AirlinesUnion Oil Company
United Airlines
Weyerhaeuser Company
Burlington Northern
New York Life Insurance Co.
Owens-Illinois
RCA IndianapolisFirst National Bank of Boston
Woodward & Lothrop
Caterpillar Tractor CompanyDepartment of National Defence - Canada
National Revenue, Taxation - Canada
Defense Intelligence Agency
Shell-Canada
Manufacturers National Bank of Detroit
Crocker National Bank
ALMSA - U.S. Army
Delaware Data Services, Ltd.
Anheuser BuschLockheed Missiles & Space Co.
Southwestern BellSinger Information ServicesXerox Corporation
Whirlpool Corporation
Software International S.A.Internal Revenue ServiceHaskins & Sells
Massachusetts Blue Cross