The Association of System

Performance Professionals

The Computer Measurement Group, commonly called CMG, is a not for profit, worldwide organization of data processing professionals committed to the
measurement and management of computer systems. CMG members are primarily concerned with performance evaluation of existing systems to maximize
performance (eg. response time, throughput, etc.) and with capacity management where planned enhancements to existing systems or the design of new
systems are evaluated to find the necessary resources required to provide adequate performance at a reasonable cost.

This paper was originally published in the Proceedings of the Computer Measurement Group’s 1982 International Conference.

For more information on CMG please visit http://www.cmg.org

Copyright Notice and License

Copyright 1982 by The Computer Measurement Group, Inc. All Rights Reserved. Published by The Computer Measurement Group, Inc. (CMG), a non-profit
Illinois membership corporation. Permission to reprint in whole or in any part may be granted for educational and scientific purposes upon written application to
the Editor, CMG Headquarters, 151 Fries Mill Road, Suite 104, Turnersville , NJ 08012.

BY DOWNLOADING THIS PUBLICATION, YOU ACKNOWLEDGE THAT YOU HAVE READ, UNDERSTOOD AND AGREE TO BE BOUND BY THE
FOLLOWING TERMS AND CONDITIONS:

License: CMG hereby grants you a nonexclusive, nontransferable right to download this publication from the CMG Web site for personal use on a single
computer owned, leased or otherwise controlled by you. In the event that the computer becomes dysfunctional, such that you are unable to access the
publication, you may transfer the publication to another single computer, provided that it is removed from the computer from which it is transferred and its use
on the replacement computer otherwise complies with the terms of this Copyright Notice and License.

Concurrent use on two or more computers or on a network is not allowed.

Copyright: No part of this publication or electronic file may be reproduced or transmitted in any form to anyone else, including transmittal by e-mail, by file
transfer protocol (FTP), or by being made part of a network-accessible system, without the prior written permission of CMG. You may not merge, adapt,
translate, modify, rent, lease, sell, sublicense, assign or otherwise transfer the publication, or remove any proprietary notice or label appearing on the
publication.

Disclaimer; Limitation of Liability: The ideas and concepts set forth in this publication are solely those of the respective authors, and not of CMG, and CMG
does not endorse, approve, guarantee or otherwise certify any such ideas or concepts in any application or usage. CMG assumes no responsibility or liability
in connection with the use or misuse of the publication or electronic file. CMG makes no warranty or representation that the electronic file will be free from
errors, viruses, worms or other elements or codes that manifest contaminating or destructive properties, and it expressly disclaims liability arising from such
errors, elements or codes.

General: CMG reserves the right to terminate this Agreement immediately upon discovery of violation of any of its terms.
 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

A CAPACITY-COSTING METHODOLOGY

Michael S. Zambruski
Boeing Computer Services Company

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
ABSTRACT

This paper describes a hybrid technical-finaneial procedure for allocating eost in a large data center.
It is designed as a modeling tool to permit "WHAT-IF" analyses for various configurations, levels of
costing, etc. The use of modular programming, independent tables, and matrix output allows the modeler
maximum flexibility in allocating cost acc~rding to changing technical and financial considerations.
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

The first segment of the procedure focuses on a technique for determining primary system capacity. The
second segment generates the cost-allocation mechanism, which assesses cost to users based on their
individual impact on capacity. The entire methodology was formulated using"- ~NS Integrated Control
System (KICS), which is written in SAS (Statistical Analysis System).

I. INTRODUCTION according to the user's impact on capacity (i.e., time

of day, boundness, type of service. host configuration,
One of the key problems in developing DP costing etc.) •
systems has been to find a co.mon grou~ between
technology and 'finance such that a methodology The analysis presented in this paper deals exclusively
acceptable to both parties would arise. On the one with HVS activity, and it is developed within the
hand, technical considerations such as performance framework of the HVS Integrated Control System
tuning have led to costing algorithms which financial (HICS.). Since HICS is written in SAS••, so are the
circles have dis.issed as overly complex. On the other progra.a which form this model. (See Further
hand, the type of accounting information traditionally Considerations section for a brief discussion of the
sought by financial offices has left data center applicability for CMS.)
managers very skeptical. This paper describes a
decision-oriented procedure which hopefully makes 2. TECHNICAL CONSIDERATIONS
reasonable progress in resolving this dilemma.
In constructing a technique for defining primary system
The basic philosophy of this method identifies capacity, an initial decision was made to limit the key
flexibility, si-plicity, and usability as the ultimate diagnostics to CPU. memory, and 1/0. thIs 18 not to
criteria of success in a cost-a110cation system. The say that processing was the sole area of concern.
system must be relatively easy to modify so that it Rather, it was assumed that high-level usage of CPU,
does not beco. obsolete in the face of new .emory, and I/O in most cases corresponds to high-level
technological and econo.ic developments. It must be communications activity, DASD occupancy, etc. Further,
simple, though in a responsible sense; that is, its it was recognized that major capital asset decisions
complexity must be tempered with practical judgment. ar~ usually driven by two performance criteria;
Finally, no matter how purportedly accurate or interactive responp~ time and batch throughput, both of
co~rehensive the system Is, It must be user-oriented. which are extremely sensitive to the central processor
configuration. In short, by analyzing CPU, memory. and
In an attempt to develop such a system, it was 1/0 as diagnostics of total system activity, it
necessary to lUke many assumptions and. draw some appeared that a minimum number of resource types could
relatively controversial conclusions. These are be studied and still yield valid conclusions for
clearly discussed, however, together with the various system-level technical and financial decisions.
alternatives considered, so that the reader can
appreciate how expandable--rather than limited--the In spite of this effort at simplification, it should be
method is. noted that the .ethodology to be discussed 1s
inherently flexible, so that it can easily accommodate
The technical section consider. a capacity projection refinements to include, for example, separate segmentB
tool in which the key diagnostics are CPU (TCB) time, for network activity, DASD occupancy, and other
real (main) meaory, and 110 (EXCP) activity. pertinent aspects of performance. At the outset,
Regression analysis is used to define predictable however, the three key elements discussed above provide
relationships among these diagnostics for each of the a sound and digestible point of departure and possess a
user groups which comprise the system load (e.g •• significant degree of validity on their own.
batch, CICS, interactive. etc.). These relationships
are used to project maximum productive CPU time--and In analyzing these key diagnostics, the following
the &Ssociated memory and I/O necessary to sustain that strategy was employed:
CPU 1evel--for each of the user groups. Further
refinements are made to accommodate time of day (prime. a. determine current consumption of TCB time,
non-prime, weekends) and boundneas (CPU, memory, 1/0) main memory, and total EXCPs at the system
as they affect capacity and cost projections. At this level and at the individual user level.
point, it 1s shown how a speCific user's impact on
capacity can be determin~~.

The financial section first considers several types of .HICS is a product of Morino Associates, Inc.
system-level costs as candidates for distribution. A
variable-load/variable-cost concept is then selected to ··SAS (Statistical Analysis System) is a product of
produce a basic cost rate. An individual user's Cost the SAS Institute, Inc.
is finally calculated using this basic rate, adjusted

287

Find a CMG regional meeting near you at www.cmg.org/regions

 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

I;OMI'''UI SON 0' CI'U 11111 11'''11011

I'W(;lHI HIHflS '0
10111I
r~AX
l'IlS~
I MlJr4 'IS. 1011'.1
11111 WAil -CIOel\ I
'Is.
ll~l
U ~>l U~;

CONI ICUIIAI ION A .- rOlt lINl MUN III

II"H "!lAlIl SLJHS

'"
P[ltClHl 5IIM
,•

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
I
I
"""""
• .....
""""" ....
"""""
.....
"""".
" ." ...
............ .....
I
• 91111
I
... ....
.....
••••• " .....
.."."".".... .""" "
... .
I
I
•I..... "
.."." ... ...."
"
........
....
.. ....
. 800
........
..
" "
......
"
I
I .....
• ......
"" "
........
..... .......
"" ..
... ........
..... ....fl.
..".""
....... .....
......
.....
.....
... ....
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

• 11I0
I
I ...........
."
.......
" ........
.....
........
.....
"
.. ......
... ....
• 6110
I
• .......
I ......
....... ..."." ...
•••••
.""."...."...
..........
•••••
.....".....
"..
..".".....
...."" ......
I
...... ""
......
...... .....
"
..... .......
..."..... .....
........ ..".......
I
. 500 •I ........
.....
..... .............
....."" .... .."........." ...... ......
.......
....
• ...... ".. ..... ........". .....
I

.... ..... "."". ....

..""" ....
.........
I
• 1100
I
......
"
...... ........
....... .....
"
.......
.•"""•••• ............. ...
..... .....
..........". .. ". ..
"."
• 300
I
I
.......".""
• ....
TTTTl
TTTTl
......"..."....."...
"
..""".""".
."...
TTlTl "."
.. " "."
.... .".."."
.""""
......
..... "
I I" I

... " .......

........ ".... ......
""."" "".
I

• ..
I
I
"..
.....
"" ......
....
T11Tl
11111
III I I
"
.•••••
....
III I I
RlUlOL!
IlUIIIllI
"""
......
............".. .... ......
"."""
.....
"." 11IlT
...""""
"
"
.....
11TH
l1TlT

.......
" I 1I "
• 200
I
I .....
....... ..".
....
III II
11111
11111
•••••
..... ".....
IlllUIlB
BlilllJD
IllIllBIi
......
"
........"
...."."
I I II I
Illlllllil
unltlllI
.............
" ""
.""".
"".".
Ilnllllll
lI111mll
lI111illll

....".... ....".
••••• •••••
• ......
"".""
I
.... IIl1ll1m nlllllHl UIIIIIlIl

. "..... ""....
I I" ,

" ... "

•••••
....... .....
"" .""
.100
I
I
I
.""
......
."""" " ......
...... fUUllln
HllIUlll
0I1111111
Illlllllli
....
•••••
""" ....
. ..
""
...
UIII1II8
llUllliU
OUIIIJIl
IUlUllll
"
".
"
""" ..."."""
". "
1I1l1l1ll1
BlUlIllI
llllillUi
1lI11llUl
.. ..""".
.. .... ""
"."
"""
"
llllllllll
IItUlBIi
llllllllB
III1UUIl

HII' 101 US[ "/IX TOT uS[ "AX >0, USL HIIX >01 US! (;Al [WHY

1--------- 1 --·------1 1--·------ ] ---------1 1--------- 4 ---------1 lUlU

SYMBOL SE.RVICE SYMBOL SERVICE SYMUOL SERViCE SYMlllll 5£ 11'1 IC£

• • 81.ICB I NTERACf rR"HSAcr

FIG U R R I

b. investigate the relationships these Current consumption measurements were made from SMF.
three major resources.
'''''"8 RHF, and TSO/MON** data accessed directly from the KIes
data "base. Separate measurements were made according
c. project maximum productive
portion of wallcloek time.
TCB time as • to configuration, user group. and time period.
Configurations were distinguished according to number
of processors (UP vs. AP). essential tuning
d. project ID8xi1llUm main lletllOry and EXCPs from (b) differences~ etc. Three user groups were defined:
and (c). batch. interactive. and transaction--which included
CICS. IMS, etc. (The model will a.llow further
TeB tlme--es opposed to SRB or total CPU time*--was breakdowns; for example. separate categories for eiCS
selected as the DOst efficient measure of user and 1MS.) Four time periods~ or zones, were used:
processing ti.e for both capacity and costing
purposes. (For example. the ratio of TCB to SRB zone 1 0800-1800
produced a relevant insight into a user's I/O zone 2 1800-2400
boundness--a key technical concept to be applied later zone 3 2400-0800
in the financial section.) Real memory usage was zone 4 • weekends.
calculated at the step level: average working set size
(expressed in pages) multiplied by the corresponding Extensive statistical analysis was then made in order
residency time. This yielded a two-dimensional to isolate and verify predictable relationships among
variable (page-seconds) which was very effective in the diagnostics. For example, the SAS procedure PROe
portraying finite Demory usage under actual operating SYSREG was used to develop independent and simultaneous
conditions. Total EXCPs were used as the most regression equations for the following combinations of
consistent indicator of user I/O intensity. Included variab les:
were measurements of disk. tape, mass storage, unit
record~ communications, and virtual I/O activity. main memory vs. TeB seconds
total EXCPs vs. rCB seconds
main memory vs. total EXCPs

*TCB • Task Control Block time; SRB ~ Service Request **TSO/MON is a product of Mortno Associat~s, Inc.
Block time. 'TCB+SRB' is commonly referred to as
Itotal CPU time' for a program (obviously excluding
operating system CPU time. or loverhead').

288

Find a CMG regional meeting near you at www.cmg.org/regions

 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

(;lIHI·"HISflH lit C:I'U III It ""110" HA:o:IHUM YS. lnlM YS. U:;I US
..llcelN' UltEAS III 10IAI rosslllt[ WALI-(;IOC!': 11I-lE
PHIMI 11M' I 060u-H!inU I fUll ON[ HOHTII
(;ONI ICU'<AlION A

l~tI CIlAJ< T 01 SliMS

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
Plllef.HI SliM

•l
I ".. ...
.
". " .. ...... ...... ......
.... ....•• ....
. 9UO t "" .. " ..
t
I::
....
::
""
::
."
::
.. "
:: ::
.
...... ..••..
.. ........•
. DUll t •• •• ".. •
I ••

·..
I "..
••

.. ". ..
• ..
...
...."* ........ ••
.. ..
••
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

I ." "..
".•• ••
.." "
•• •
..
.ICKI
"
. . .... ."..
1 .... ... .. .. ".. ."

• 6CHl ·..
I
I
... ..
••
•• ...
•• ••
.. ..
••
...
•• ••

..
...
• •••
• .
.* ..
"
"
••
"
,.,
.......
"
..
..II
II

...**.. •..."
••
.••...*.
••

...".. ..."
I •• •• •• •• •• • ••• ... lO .. II" •• .. • • 11
I •• ... •• •• •• •• • ••" •••• "
• ~f10 t
J ••
•• ••
1 •• ••
•• ••
... ••
•• ••
." ••
•• ...
• ••"
.... ••
." ."
•• ••
... ••
•• ••
... ....
• •••
•
•
.
. .. ." ...
"••II"*
lOll

*..
." ... "

...... ..
I ... •• •• ". ... ... •• •• •• ••
.
• ••• ". "*
II. II" .. II
I •• •• • .. ".
"
•• ". •• •• •• •• • ••• •• •• II" ." "II "*
II" ••

• 1100
1 •• ••
I ... ••
l :: ::
...
.."
:: ::
•• ••
...
:: ::
II. •....
•• ••
11
:: ::~: ::::;:
• ••"
•••• 11
." ••

::
•• fl. fl. ""... .•••
•• •• ... ••

:: :: n... 11::" :: ••
•• •• ......

• ]00
1::::11 ::::n
I •••• 1 I •••• II
::::~;
. . . . . II
.... :: *"::::
• • • • II ...... , I
···"~1
..... , I
::::n "·::n ::::1: ::::11T'
• • • • II ..... , , • • • • II ••••
.200 • • • • • 11 •••• 11 •••• 11 •• 11 •••• 11 •••· 1 1 · · · · 1 1 · · · " 1 1 ···"11 ·"··11
I . . . ." II •• II •• II ... II ...... II ....... II . . . ." I I · · II II .... II
I •••• I I •••• I I • • • • II • • • • II • • • • II • • • • II • • • • II • • • • II • • • • II •••• I I
1 ••••
.100 . . . .
I
00
DO
80
•••• 11
II. on
iiI)
•••• 11
.....
II
••
80
. . . . . 11
••
••••
08·"
• .... 00.···11.···11

B8 . . . . .
DB
DB
•• II
• • • • DB
II
8ft . . . . .
····11 . . . . . 11
I I·" II
DB "" •• DO . " ....
····'1
00
on
I " 09 un ". "" BD ." •• 06 • • • • DB . . . . . IlII • • • • un ."." DlJ ". ". 00 . " " . liB
I " .. Oil "" 80 ....... 80 ..... 80 ... Oll O D " " liD"" 00 Oll·" DB
--_._------------------_.------------------------------_
HTU MIU MTU "TU HIU MTU
.. _----------_._---------------------------------------
MTU "TU MTU MTU
AOS
XT[
1-- 8 -I
ADS
XIE
1-- 9 -I
AOS
XTE

'0
AOS
XfE

"
","os
XTE
IZ
AOS
XTE
13
AOS
xTE
AOS
XTE

"
AOS
XTE
,. AOS
XTt:
17 1I0UA

SYMBOL SERViCE SYM80L SfRVICE SYMUOt S£RV I CE SYMiSOL S£RVICE

• • • BAIClt INtERACT T TRANSACT

FIGURE 2

Various other SAS PROCs (e.g., REG, UNIVARIATE. RANK. categories, each of which expressed overall CPU tl111e as
PRINCOMP, Ii. RSQUARE) were employed to analyze a percentage of total possible vallclock ti.e for that
residuals, test for collinear1ty, and verify the zone:
rationality of variable combinations chosen. The goal
was to predict the level of memory and I/O activity a. MAX - total time for which the CPU .aa
which would be necessary to 8ustain a particular level avaUable.
of CPU activity--both at the system level and at the
user-group level. For this purpose, it was found that b. TOt - total time during which the CPU was
the simultaneous regression equations provided the best actually executing instructions.
solutions for interpreting the independent and mutual
variability among the diagnostics. c. USE - total time attributed to executing user
progra.. (i.e.. 'productive' ~lme). as
The next step was to project . .xlmum CPU time. Since oppoaed to system overhead activity.
commands are processed non-concurrently within each This was further subdivided into the
processor. it was decided that maximum CPU time would three user groups--batch. interactive.
be defined in terma of percentage wallclock ti.e. This and transactlon.
would provide a fraw of reference which 1s technics tly
reasonable and yet readily acceptable to a In atte.,ting to project . .xi~. productive capacity,
non-teehnically oriented audience. such as finance or the following logic was applied to Figure 1. The
management. Also. this fundamental concept of a difference between the MAX and TOT columns represents
CPU/vallclock relationship would allow direct tote1 unused capacity; the difference between the TOT
measurement (rather than a standard capture ratio) to and USE colu1ID8 represents system overhead. Based on
be used in finding 'productive' capacity--l.e •• that this, we can grow the TOT columna to aome level at or
non-overhead portion of system capacity which is below MAX to signify maximum practical CPU. and then
available to users for batch. interactive. aDd grow the USE columns to show the corresponding ..ximum
transaction work. productive CPU. In this regard. it was assumed that a
basically linear relationship exists between TOT and
The approach taken to this task is depicted in USE. 80 that any growth projected for TOT would yield
Figure 1. which shows overall CPU activity (TCB. SRB. linearly proportional growth in USE. To teat this
and system overhead) for Configuration A. In this assumption. zone 1 was profiled for an entire month on
case, individual time zones were broken down into three an hour-by-hour basis (Figure 2). This gsve more

289

Find a CMG regional meeting near you at www.cmg.org/regions

 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

CONCEPT OF -RESEAY(O CAPACITY-

(WHERE GROWTH IS PROJECTED FROM CUftREHT LOAD'

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
Oyor.11 Projection -Re,erY9d- by seryice
MI. 1_ Uur Leyel
(_IS or c8p.cltV)
- • .-~~-~~-~~-~-~---.
I
I• I
•
I I
I
•
I II TRANSACTION
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

Current User Level ..

CUI or c.paCI~.
-
TRANSACTION

:/ INTERACT lYE

UtfERACTIVE

BATCH
. .TCIt

PIGUa. 3

visibility to average peak periods and their associated by de.illl-the workload mix. chan,.. dra1l8t.ically. then
TOT-USE profUes. Thf! ruu1t8 appeared to support the the -.odel Deed only be chanled to reflect the new
".U1IIptlon of b..lc linearity. Accordilllly, tbe TOT relerved capacit.y for these subgroup., and the process
columut in Pigul'e 1 were &r0Wll to the 90% level (above will continue accoTdingly. 'Ibis .ekes i t possible to
which aoy ll1le8.l' pred1ct1008 would pl'obably be play 'WHAT_IF' . . . . wit.h var1ou8 workload types. ti_
invalid), and the USE columns were increased dist.ributions. et.c.
proportionately. Thus, vheu TOT for zone 1 was grown
from 70% to 90%, USE for that zone vas incl'ea&ed fr01ll At thle point. the maxi... pl'oductive CPU capacity
331 to 431: rellerved for each subgroup and zone was translated int.o
Tea ..cond,s. 'nile waa done by cQ1IParing TeB time O8a.e
(.9/.7)*(.33)-.43. vith total CPU time usage by subgroup and ~oo.e. and
then applying these actual ratios to the projected
Each of the USE colulIIb8 were sill1larly increased. with capacities (which were expressed .. perc::entages of
the results representing the maximum productive CPU wallclock ti.e). The resulting new percentages were
capacity for each zone. multiplied by total pos.ible valle lock time to yield
TCB seconds. All that re_ined was to U8e the
Because the USE columna are subdivided into the three regression equat.ions discussed earlier t.o est.ablish t.he
user groups, the overall capacity projections had to be _xi1lUlll main mentory and EXCP levels associated with t.he
prorated a1llOng theae subgroup.. Figure 3 outlines this nevly-defined aaximum productive TeB times. A
concept, tented 'reserved. capacity'. In this case. i t three-di.ensiona1 profile of productive syste. capacity
was decided to project a separate maximu. for each vas thus complete.
subgroup in relation to its exi8ting impact on the
syst.em. In essence. this . . . .eel that the future It was now possible to establish a specific user's
stat.us of the8e 8ubgroup. would reflect the present impact on productive system capacity by comparing
status. Bowever. should i t happen that-by aCcident. or ~..ured resource consumpt.ion with t.he ealculat.ed
capacity. Figure 4 i. one exa~le of how t.his can be

290

Find a CMG regional meeting near you at www.cmg.org/regions

 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

NYS CAPACITY IMPACT PROfiLE

USER )( fOR OftE MOMl H

-- ••..----- (lj (Bj--- (c)..[n)-.- SYSll~E'flN·IF\CA·, . . ·tc:i-·----(-Hf---····(i·f--·····(j)·····-(k)--·····(i.)·-

T M M ME E E
S 1 C T [E HE )( )( M )(

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
E c B MC M M AM C CAe
II D' AB Y P )( Y P P X P
Y Z LA JC lAM LA E
I 00 R T~ 0 R [ M 0 R"': it
C "A T CA A 1 H A AT C A
[ [0 LOX 0 L V X D L ' X

DAICn
BAICn
1
2
9,"88
l,6JJ
.U?
•1I'.i I
~":J,J62
316, J611
.011
. (1')11
14.9 ,1,"11
'
a. 71~,6~1
."'5
.085
l,lJ6?,613,261
'11'1,78'" l!i6
, (H,.
, (118
J,161,'106
1,19':0.530
• J~ I
, US
11l,12/,14m
105,109,1161
.U211
.1111
DAICn
PATen •
0,
1
9!i1
12,8117 •
.031t
. nOli
.'1511
S"It,171
!lo18.952
116,"98
.002
.000
.013
5o,1SI,U2
196.136
5o.092.2fio7
.050
.n02
.0119
686,188,803
729.096.S10
1lt1.963,832
,(111)
,11110
,lU'l
953,7lt9
'l,l?S
2,8%,1';0'1
• Ilia
.0110
.322
153,0119,929
1(,;/,. 611). 2(j1)
'13.216.276
.U06
• nOli
.066
I"([RACT
••••• .Ul')
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

INJ[tlACr 2 .fl21 3".811 .017 JS7.279 .003 23.363,3113 91.139 .010 3.620.83') ,fJ2!.1
UU[RACT
INI[RACT •
"
1
.nCll
.09]
11,102
17,91"
1!t,81 1
.00.
.167
281.1I20
65,916.8111
.on]
. (,39
22.080,97.3
31,111.'180
'56.101,S"7
.UU
.185
21,689
lIS.8;!9
.0112

.Ol6
3.88'1.327
2,7 ,,1.62
1.7611.161",
.006
.u8 11
TRANSAct 2.6116 '
9,0')8 .u1'J2 3')J, ')"1 .00] ]11.5'115.618 J,~IJ6 8.9'UJ,IUft .0rlO
,
lRAlISACr 2 15 .0411 .010 .OlllJ
TRANSACT 3 2.' .'MII) 3,"38 .078 1.691.ftIO .fl1fi ISIl.6"17.998 .011 2S7.0'.8 .0?9 9?'I,tlB6 .278
TRANSACT
• .1)011
.... -._--
3,162 .001
... _._._---
425,'ISfio .00' 159.161,326
.3,878.]03.851
__ ._--------
.001
'"
---------
.(Jf,O '185,907
-----------
.001
SVSIO 28.5116 2,316.610 103,2118,961 8.861,213 620,0112,096

. __ .---------_ ... _.. __.. _... _------------------------ SVSTEM lo["rlrICATIOH~D

-_._--------------------------------------------------
, • , M M M , , E
S
E c •
C
M C
l
M
E
N
M
A
l
N • C •
C
M
A •
C
R • •• ,cx ii•
A Y
• • y
• • x •
, ,• ,••
V l l l M l l
0
•• • N
• R E ii 0 x ii
c • • ••
A M A A C A
E
eATen
BAICII
E
1
2
21.0119
8,9'5
0
.... l

.o;?'}
18).606
281, :lS3
.113
.032
16,966,117,
22,751,310
0
.031
.0'"
l
326,Z39.963
60f4,909.8"3
y X

.O~2
.038
3,028.911
].278.916
0 l
.061
.066
115. HU.9fJO
92.283.920
• • .067
.036
DAICIt
BAICIt
INI[RACT
3
•
1
12.233
IIS,'II'
20.1&78 ....,
.0110
.11&9
'116,63"
526.2'12
309,0117 .....
.029
.08'
15,910.912
'7.317,811
7.326,337
.029
.103
.013
597,232,168
753,877 •.1(31
1311,886.609
.027
.016
.05"
II. 732. 39'i
a.527. IDO
1,117,158 '
.095
.112
,063
11",990.9"19
1115.21'2,805
69,012,';069
.0'11
,1")'9
.0115
INrERACT
INI[RACT
INTERACT
lRA11SACf
TRAIISACI
2
3
•
1
2
2.212
2.361
2.376
60.625
25.9.Jl
....
.00'
.flfl8

...,
. 199
611,222
39.i167
6J.168
132,185
'11.608
.03_
.06.
.037
."59
.7..9
995.108
1,679,01,
687,110
JOO,7911,078
112,130,371
.002
.003
.001
.5"3
.077
112."23.000
23.1"8.88"
12,1"",1I5S
1. 166.831!1.673
'1,061,39"
.023
.073
.021
.258
.7,.9
3118,11111
'2",019
519,697
10,920,953
",981,1199
.007
.011
.010
.220
• HID
9,211,578
10,31.1,03]
13 .1110.965
21,113,3111
6,658,735
.rU8
.051
.039
.517
.1'18
llW1SACT 3 22.189 .Oll 5oS.n7 • 1'7 111#~nl.1112 .02• III&S,855,3IS ,011 2,009,032 .0'12 1,129.713 .291
TRANSACT
SYSID
• 10,95"
301l,9f(2
.265 156.113
------_.-
2,26S.2a3
.519 12,132,296
-----------
5053.169.299
.130 1I1a,123,013
-------------
1I.602.7JtO,887
.113 l,50ltl,316
-.--------
119,609,52S
.1')2 111.9"",2 13
------------- '
511'.lI15,7SIt
.505

==........ ==.:=,.==="'== "":0:"'''''''''''':''''''=''': :........"'=== =============

Jl] . . . . .
• ",581.'95 651.118.261 2 0."81.01lIt,71t 1
'
~a,"76.73a 2 1,169.517.850

1 IGURI 4

portrayed. The ..trix forwat de.oaatratea the 3. lIRAIICIAL CONSIDEIlATIONS

vbibiUty cd control whieh the WJde 1 at forde :in
aoalyziag an individual user: As Mntioned early in the laBt a.etion. CPU, -.:»ry,
and I/O are evaluated as duanolticl of total aystelll
a. leparate atatisties are .bovn for each activity in tbis _tbodology. According,ly, overall
configuration (.y.t_ identification). aervice cost.--rather than .ere proc...ing or mainfr.-e
type. and tiM fra. (zone). co.ts--were used 88 the basis for coat allocation.
Three diatribution options were considered:
b. the uaer's load -.ounta (COlu-n8 A, E, t) are ayatee-incremental, capacity-everapd, and current-load
direct __ ur~nt. of the key diagnoa tice (Figure 5).
uaed in tbe _thodoloIY_ The .. aociated
I_PAllTL' frac::tlona (col. B, F. J) represent • 10 the .yate~increIDental option. base11ne coata are
.ervlce-by-aone breakdown of tbe load within allocated only to the exi8ting workload, vberea.
each confiauratioo. ay.te~level delta coste are pa.sed on only to future
u.era. The idea i8 to ..sess the I118rainal impact of
c. The HAXTCB, MAXKEMY. and HAXEXCP aaounta upgrades, etc., on the potential u.er. The problem ie
(col. C. G. K) are tbe ayatea's productive that true coat ca.parability (past to preaent) and data
-.axi.. .. deterained by the r ..erved capacity baee accuracy becoae questionable.
concept di8cuaaed earlier.
The capacity-avera.ed alternative uDiforaly allocatee
d. the 'MAX'
variabl.. (col. D. H, L) depict both baseline and 1ncre1lental coata. It nates that
the u~r'. fractional impact on the maxiu maximum utilization and coats will be eati..ted at the
defined above in (c). beginning of a period (e.g. I a year). and all usage
during that p@riod viII he casted at a constant rate.
Thi. ..trix will be used :in the next eection .. the If the predictions for the period are very accurate.
basis for tbe C08tiag vehicle. this method can be III08t desirab Ie. However. due to the

291

Find a CMG regional meeting near you at www.cmg.org/regions

 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

Cos t -{4l 10 C at on ~le Hi 0 ds

15

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
- svs-ur.:
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

C,:<'-A<E

CURLDADI
--- CURLDAD2

• 1 , • 5 , 1
RESOURCES

PIGURE 5

nature of capttal uset p lannins, i t is very likely from the cost distribution systea. Such a change would
that the anticipated levels of COBt and utilization require only minor effort to imple.ent.
will not coincide as intended. To the extent that this
disparity exist•• all usage during the period will have Continulnl with the eurrent-load eonc:ept. total pre.ent
been under- or over-costed. and retroactive adjustments .ystem CO$t (e) is div1ded by total present user load
will have to be considered. Also. any interim (T) to produee a basic usage cost rate (a), expressed
Il4nagement decislons which 'Were made may have to be as dollars per TeB second:
reexamined.
CiT • R.
The current-load option a180 encQGPasses fixed and
lncre_otal costs. However. in this case all If a user 1& not CPU-, .emory-. or l/Oo-bound. then biB
distribution 1s made according to the prevailing user 1ndividual cost (c) is determined by his TeB load (t)
load and systeJll costs. This means that an individual eultlplied by the basie coet rate:
user's costs are deterained not only by his workload.
but also by the workloads of all the other ueers on the c • t " R.
system. As the total load changes, each user is
affected. The salle applies to the system costs: each However. 1f the user is resource-bound in one of the
incremental cbange in the system cost profile affects three key diagnostic are.. , a pretium (p) 10
each user to some extent. The idea here 1s that true established which increases his total coat by
cost absorption is a non-static Bituatlon. Att8lDptS increasing the basic cost rate:
(like those cited above) to simplify the distribution
mechanism by using a fixed rate or so.e limited portion c • t * (R.*p).
of total cost can potentially confuse ~re than clarify
the issues. In a large data center with heavy user The magni tude of the preaiua depends on bow auch tbe
activity, the current-load method offers a dynamic user's boundness adversely impacts cap.eity. An
solution to 8 dynamic problem. For this reason, it was example of how this boundne&s is determined and then
used to generate the basic cost rate described below. translated into a coat premium is shown in Figure 6 and
described below.
Before proceeding, it is appropriate to mention that
any of the above three allocation alternatives can be The basic structure of the preaium table derives from
used with the model. In fact, depending on the purpose tbe capacity matrix described earlier--viz.. a
of the analysis, it may be logical to make separate breakdown of the three key diagnostics by
runs for each method and then compare the results. configuration~ service type, and 't1_ frame. The
Also, the model can accommodate other distribution '-PARTH' variables (col. B, E~ H) are 'tied to the
concepts. For example, the three options just reserved capacity concept outlined in the prior
discussed share a common premise: All costs are section: each value refers to that part of 1Il8Xi.-1l
completely applied to productive capacity, with no productive capacity which is allocated or reserved for
costs separately apportioned to system overhead the associated service and zone within the
activity; therefore~ each user i1llP11citly shares in configu~ation. Fo~ example. column B for Config. A
absorbing the cost of system overhead. However. the shows that 23.5% of the total productive TCB capacity
methodology allows the modeler to change this premise for that configuration is allocated to prime time
and address system overhead as a separate entity. apart (zone 1) batch service. Another 16.2% is allocated for

292
Find a CMG regional meeting near you at www.cmg.org/regions
 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

CALCIlLATION OF f'1I[HIIIH US£O 1ft CAPACITY-COSTINe

USER X FOR ONE MONTH

J K L
-------------------A----*---B----~---c----·-----D--- :VSl ~ I n[NII F liAr InH~A 'G-----*-----i----,;---i-------(F7c)-----Ciicr--Ti-ir
S£RVICf lONI TCB_MAX TCBPARTM lCBFI\CTR "[MY_MAX M[MPARTM M[MFACTR [XC::P_HAX [XCf'ARTH XCPFACTR HEWi_TCO [X(:,_TCB PREMIUM

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
1 .1117 .2l') .(J(III .11111 .2l 11 .IMII. .02 .?11 .no~ 1).f)11 1.2!:> 1.17
IJAIl:ll
.lm
tlAf(:1I
'IA'CU
QAICII
,
2 • (Jell,
• flU?'
.000
.14>7
.2]5
.2'.10
.IMII
.0011
• OliO
.U18
.fl06
.000
.122
.171
.188
.,tcl?
.(M)1
.f100
.011"
.oM
.lllJU
.?',7
.262
• flO?'
• (IU?
.lInll
].21
J.n
111.8
2. III
3.13
1.9/1
8.18
12
28.6
'I .1110 .1105 0.211 0.83 0.20
IHI[RAe, I .013 .UlG • illiG .03 11 .(138 .f1Ul .1166
I NI £llA(;l 2 .017 .111') .OllfJ .Of'} .0116 .11110 .u?'} .UII(, .fllIll 0.3(, O.~8 0.21
INIUIACI 3 .(J08 .UIl') .4Jlm .013 .Ufl6
.01l8
.Ollll .00(, .11116
.IIU II
.IUJlI 1. f)fl 0.9? 1.'"
INlrllA.CT .11118
UtANSAel "I .167 . fill' .Illil .185 .092 .011 .08 11
.Of~
.11116
.fll
.001
.IIUl;
111.9
111. 1
1l.115
1l.89
6.611
12.1
IIIANSACI 2 .002 .111111 .UflO .010 .1109 .fIIllI
1RAN SAC I 3 .018 .f1UI .1100 .flll .1139 .IIUO .U8 .IIUl" .1I0U 3.1l. 3.51 13.'1
.ouo
• .0 111 .001 .0rl1 50.2 O.3? 16.2
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

lltAHSACT .001 .lIfI? .000 .003 .000

SYSID

..
--~--------,_._-------------~.---_.- ----_.~-----_.- $YSIIM II>I.NII r 1(;" liON =11 ---.----------------------------~---------------------

SERVICE z.., len_MAX f(;nf'AllTM TCllfA.crlt MEMY_MAX MlMPABIM MLMfA(:llt [XC,._MAX [XC;f'''''I" )(CPfAr.llt "I HY_Ten [XCf'_'c:n 'lltHIUM

DAICII I .113 .1182 .009 .O'}2 .011 .nu', .fl61 .082 • flOG 1I. l lll 0.')9 0.2 11
.• 032 .1211 .flOlj .038 . III .UI)") .036 .1(.3 .(m6 1, 25 I. ~o 1.88
BA1CII 2
BAH:U 3 .li29 .18 11 .005 .027 .1311 .Otl] .0111 .209 .U1l9 0.6 11 1. 59 1.11~

BAlal
IHIEIlACl •I
.0'"
.066
."32
. 13(,
. 020
• Of)'}
.OU•
.U")II
• 1611
.02')
.012
.(J1l2
.O~9
.u ,5
.26'l
• 1?6
.016
.OU6
0.62
0.18
fl. 71
0.63
0.118
0.11
'HIEIIACJ
INI[IlACI
IHIUIAl::1
,
2 .03"
.061i
.031
.fJ28
.017
.ole
.1101
.001
.IMII
.023
.013
.021
.!J09
.005
.lJlH
.(mU
.fJOU
.(1011
'
.038
.051
.Ol9
.1111
.Cl19
.025
. fill I
.1l01
.001
0.22
(I. l~
li.tI,
f1.65
(J. 91
tI.911
0.111
0.32
U.1J
TRANSACI "
1 .459 .0'.i8 .027 .258 .254 .065 .511 .1I38 ,020 2.fl l l 0./11 1.81
1 MHSAf: I
TRANSACT ,
2 .7 1'9
.391
.015
.025
.011
.010
.7'19
.033
.1112
.091
.009
.00'
.7'18
.293
.fl12
.013
.on
.(14'9
.001i
0.81
0.32
0.19
0.39
0.611
0.12
TRANSACl

SYSID
• .519 .fl69 .ol6 .113 .091 .016 . 50'} .011• O.li't 0.l8 0.11

2 2 2

P I GU IE 6

zone 2 batch service, and 80 on. ObvIously, each of The premium table perfor-s this analysis for each
the '-PARTM' columns adds to 1.0 for eaeh service type and zone of each confixu'fation. The
configuration. the '_MAX' variables (col. A, D, G) MEMFACTls and XCPFACTlls are compared. to the TCBFAClas
represent fractional impact on the capacity allocated in each case (eoluans J and K) to determine whether
by the '-PARTH.' variablefl. there 18 MlIOry-boundness or I/O-bouodnes8, or both.
If there is both, the table assumes a geometric rather
The purpose of the I_PARTM I variables is to normalize than arithmetic relationship and multiplies the
the I MAX' variables in order to ascertain the true comparieon factors together (colu-a L). The result 1a
relatlve impact on capacity. The following example a premium which is then used to increase the user's
illustrates how this process works. basic cost rate. as described earlier. In this way.
any un~ual i.,act on capacity (due to boundne••) 1.
Colu.n 8 for Config. A shows that the productive TCB directly translated lnto higher cost.
capacity allocated for tone 1 batch service is 23.5% of
total productive TCB capacity available for the It should be noted here that the capacity amounts
configuration. Column A shows that Uaer X consumed corresponding to the '-PARTM' fractions were determined
1.7% of this allocation. The user's relative impact on by the regression analy8i8 and reserved capacity
TCB capacity here is therefore 1.7% * 23.5%, or usumptioD8 aentioned earlier. One of the primary
.004--86 shown in column C, called the TCBFAClR. If virtues of this methodology is the ability to change
the same calculations are performed for columns G and these basic alDOunts as experience and/or corporate
H, we find that the user1s relative impact on I/O strategy dietatee. 1f, for example. it is decided to
capacity for this zone and 5~rvice i6 .005 (column II allocate .are of the existing system's capacity for the
or XCPFACTR). By comparing the XCPFAClR to the TRANSAct vorkload and 1.s for the INTERACT work.load,
TCBPACTR, we conclude that the relative impact on I/O the .ade! need only be changed and the premu. tab les
capacity is 125% of the relative impact on TeB capacity wUI develop values bued on the new cdteria.
(coluam K).
One additional refinement is ...de to the prelliUll before
On the other hand. i f we had merely compared the I MAX I the userls cost profile i. finally generated. Thi.
values (.024 for EXCP and .017 for TeB), we would -have consists of 8 eost-weighting table (Figure 7) which
concluded that 1/0 impact was 141% (.024/.017) of TeB double-ch~ks whet.her t.he pre_ium 1a rat.ional for t.he
impact. This would have been an oversimplification. time of day when the serviee is performed. For
since it. would have failed to consider t.hat the I/O example. Figure 6 llhows a premium of 28.6 fDr weebnd
allocation (colu~ H) is actually smaller than the TeB (zone 4) BATCH work. on Confi8uration A. This is not
allocat.ion (column B) for this zone and service. Using real1stic. 8ince week.ends are traditionally low-load
the '-PARTH' values t.akes this into consideration and periods. 8s8e881ng a cost pre_b.aa of 29 to lIuch work
thereby normalizes the impacts initially suggested by would unnecessarily skew the true capacity i~act of
the 'MAX' values. It thus provides a much more that workload. Consequently, the ~ost-weighting table
realistic portrayal of whether a us~r 1s 'bound' in a allows the modeler tD override the premium. In this
particular direction of resource consumption. and to
what relative degree.

293

Find a CMG regional meeting near you at www.cmg.org/regions

 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

: THE fOLLOWING TABLE: ACCl:prS OR OVrRRIOES THE PREMIUMS APPLICABLE

** TO THE BASIC T C BOO l It
.,,,,,.,,.,,, RATE (FOR INDIVIDUAL USER COST).

**--~~~----~----------------._--------------------_._--._._._-------

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
SYSIDI ZONE 1 I ZONE 2 I ZONE 3 I lONE II
*-~----+-~------_._------+---------.-------+---~--------+-
* I I I I
...------
* IIC8LOAO*TCBRAH I I I
* I I I I
* I( IF CPU-BOUND I I I I
* I I I I
I I I I
• I -OR- I rCDLOAD * lCBRAT[ * 0.5 I TC8tOAO*
• A I , , , TCBRATE*
• I I I I 0.2'5
• I TCRLOAD*rC8RAI£* I I I
• I PREMIUM I I I
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

I I I I
* 1(lf 1/0- OR MEM....-I I I
* I BOUND) I I I
* I I I I
* I I I I
*--~---+---_.--~------~--+-._---~------~---+---_.~------+----------
I ~"''''====:="====::=,, I I
• II NTERACT I I
• ,
ITcBLoAD*rcBHATr* 1
., , I
I
• I 1.3 I I,
• ,
III F CPU-BOUND.
I
I :
:
*
I
II -OR-
II SAMI: AS ZONE 1\
II
I
I {)
*8 I SAHrAS I SAME AS
* I I rOR COMfIG. A CONrIG. I COHfIG.
* I TC8LOAO*TCBRATE* I A I A
* I PRE"'tUM*1.3 I I
* I I I
* 1(lrl/O-ORH(M....-1 I
• I BOUND) I I
• I I I
• 1===='""'''''''====''''''=== I I
• 18ATCH II TRANSACT I I
• I====="''''=,,==='''''== I I
* I I I
* I SAME AS FOR I I
* 1 CONr IC. A I I
*--~~------_.-._----------------_._~------~--_._-_. __.-------------

I'IGU"R. 7

eaae 9 Figure 1 shows that zone 4 work 1s uniformly 4. FURTHER CONSIDERATIONS

1d1acounted' at the rate of 75%, and the pre1ll1uIlI8 are
ignored. Zones 2 and 3 are treated 1n much the same ObviOUllly. the preceding HctioM were limited to a
way. Zone 1 ia treated differently for the two brief overview of the ba.ic .ethodolo,y. "-ny
configurations since they are tuned dlfferently--i.e., additional appl1eationa and refin_ots are poeaible 9
Configuration B is deaigned to provide better response aeveral of which are described below.
time for interactive work. Accordingly. INTERACT
aetivity is assessed a 30% additional 'penalty' on top lb.e matrix approach to capacity-eo8ting sets the stage
of any appropriate premium. Once again. the8e choices for a very powerful type of -adeling: optimization.
are entirely up to the modeler. Any chanle. made viII Here the goal ia to find the particular
carry through the reu.ininl procenlDI which the . .del cost-dlstribution matrix whlch opti.tze8 both capacity
perfonu. availabUity and cost recovery in one or IIOre contexts
(e.a·9 zone. service. configuration. key diagnostic.
The final result of this process is a user cost etc.). The flexibility of the basic methodology makes
profile 9 where cost impact is a function of capacity it practical to perfora auch a study both at the system
i~act (Pigure 8). The familiar ..trix format shows level and at the individual user level.
user load and effect on capacity in the three key
diagnostic areas BrrBnsed by configuration 9 service. One area not addressed 1n the prior sections concerns
and. zone. lb.e pre1ll1u. are indicated (although they user-specific costs. An example ia a library of tapes
are overridden in certain cases as outlined above) whieh is available to only one user. Coats a8aocla~ed
together with the resultant cost calculations. The with such a library should only be pa••ed on to the
last eoluDn. OOSTPART 9 allows the user to see how the party involved and not to the general uaer base.
model i8 ultimately distributing costa. In this Another example would be a Bdnlmu. cost allocation for
exa~le. the majority (37.1%) of this user's coat is 8 . . 11 users 9 whose impact on capacity is minor but who
co.ing from zone I TRANSACT activity on Configurstion nevertheless are responsible for significant
B. The .adel will generate this type of output for administrative coats simply by virtue of being on the
each user and for the entire system. thus allowing system. In both of these case8 9 the generic proces8
total Visibility and control over the outcome of all outlined earlier can easily be tailored to .eet these
the assumptions and decisions made up to this point. needs by the addition of further tables.
Higher level su.maries by zone and service (Figures 9
and 10) then become a simple task which nevertheless An area of particular interest to technology would be
produces very informative results. the accuracy of the maximum amounts derived by
regression analysis (Figure 4. col. C. G, K). Testing

294

Find a CMG regional meeting near you at www.cmg.org/regions

 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

MYS CAPACIIY-COSTING PROfiLE

USER X roR ON£ MO.U II

COST TCOLOAO • OAS{RAl[ • PR[MIUH(OR OV(R'tlO[)

IOfN TI r 1CAl loN=A ----------- ----- --- ------- ----- ------------ -----------

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
SVSUM
-----------------------------------------------------
SERVIC£
'ott' TC8LOAO TCD_MAX "["nOM
1'1.9 117,'111
M£.MV_HAX [XCPLOAO [XCP_MAX PR[MIUM COST COSTI'ART

•,
BAICII 1 9,1188 .011 .014 3.167,Qfl6 .112" 1.17 11.8 1W .038
DAfCII 1.633 .uo,. 8,775,657 .018 1.195,530 .011 8.78 a.a .1103
BA1CII 957 .002 5,151,332 .OU8 951.7'19 .006 '09 .002
DAICI!
• a .000 196,736 .UIJO 1I,12S
2,856,75 11
.000 28.6" • .000
IHIlRACT
I"I[RACT
'"IntACT
1"Ir.llACl
I
,,•
12,8'1 ,
'90
aa
.013
.017
.ooa
5,092,267
157.279
2111,"20
.03'1
.015
.013
91,139
21.689
.066
.025
.006
.197
.208
1.7'1 ".
13,659
.7
• fill 3
.001
.000

TnAHSAeT
,
1 2.6116 .161 65.916.8'17 .185 ]15,829 .OBII 6.6'1 18,685 .fl59
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

lRAJlSAC1 2 .002 353,5'11 .IHO 3,su6 .uem 12.7 a .oon

TRANSACT "
"69 .078 1,691.010 .011 251,0 118 .278 13.4 1'13 .1100
IRAtfSACT
SYSID
•28.!.i'16
> .001 '125.'156
-----------
103,2'18,961
.001
---------
8,867,213 '" .001 16.2 1

'16,075
.1100
.1 116
_____________________________________________________ SYSTEM
IDlNTlfICATION~8 ------------------------------------------------------
S[RVICE ZONE TCDLUAU TCB_MAX MEMVLOAU MEMY_MAX [XCP'DAO OCl'_MAX l'U[HIUM COSI COST PART
BArCII 1 21,0119 .113 16,966,475 .05" 3,U28,917 .061 ."3~ 22,380 .lJ71
D~lCII
DA1CII
2
, 8,985
12,2]]
.032
.029
22,151,370
15,970,912
.018
.021
3.278,916
!l.132,395
.0]6
.0 111
1.88
1.02
11.912
6.503
.OSl
.1121
BAoICII
• "5,5119 .081 57,317,811 .016 8.521,4180 .059 .1178 12.107 .018

,•,
INIERAer 1 "0,'1/8 .06. 7,326.ll7 .O~" 3.117.158 .0115 .110 "8.3U'1 .09U
'"'[RACr 2.212 .ellil 995, '08 .02] 3' 18.1'1'1
11,019
.038 • IlIII 2.351 .007
'"IERACT
INIUtACT
TRANSACT 1
2.361
2.3/6
60,625
.060
.031
.'159
1,619,019
681, 110
300,79'1,018
.013
.021
.258
52
519,697
10.920,953
.051
.039
.517
.320
.128
1.81
.12
1.255
116.892
.n04
.U02
.371

,,
TRAHSACf 2 25,931 .1'19 !l2,730,311 .7'19 4.?81,'1'}9 • 7 118 .6142 27,511 .081
lHANSACT 22, HI9 .397 '".!H8,"'2 .Oll 2.089,032 .291 ,125 11,796 .037
TftAH5ACT 80,95'1 ."9 72,132,296 .113 1,5'11,316 .505 .168 21,518 .068
SYSIO 30'1,9' 12
-----------
5053,869,299
----------
49,609.525 269.221 .8511
=========== ==========
3]].'188 651,118,261 58,"76,738 315,296

rIG U R R 8

MVS CAPACITV-COSTING PROfilE

USER X rOR ONE MONTH

SUMMARV BV TIME PERIOD (lONE)

ZONE TCBLOAO ICB_MAX M[MVLOAD M[MY_MAX [XCPLOAD [XCP_MA)( COST COSTPART

l11,U;:O .09 1111,101,4'" .13 23,1107.011 .07 211,159 .•7
2 39.366 .05 1'.963,332 .0. 9,898,13 11 .Ul, 19.0""
'
.1 •

18.u98 .0', 39,292,105 • 02 8,511,933 .u, 21J,"53 .06

4 '28,892 .10 UO, 159,1109 .0. 16,593,055 • liS 1'1.260 .11
====",,,,=,,,,,,,,,,. =""==,,=,,,,=
657,118.261 58,1116.738 315,296

FIGURE 9

MYS CAPACIJV-~OSTING PHOfllE

USUt x fOR ONE MON1H

SUMMARy BV SERViCE TYPE

S[ltVIC£ TCOLOAO TCO_MAX M[MYLOAD ME"V_M~X EXCPLOAO [XCP_ MAX COST COSTPART
OAlcn 99,901 .Ill 1..... 071. -1CI3 .f13 2'I,888.51f1 •OJ n.1211 .n
INIUtACT '111.95" .06 16,1118.539 .0 11 1,'118.601 .(l!.o 116.562 .0>
JIlAHSACl 1')<',63" .01 '198.622.018 . Ie 26.109,619 .111 1<J6.6JJ .6'
"'~="-==""'~"~ """"""'''''' -,~~
33.1,/188 657,118,261 5f1,'116, '1]8 ]1~.;>96

r I G U R R 10

295

Find a CMG regional meeting near you at www.cmg.org/regions

 Learn the basics and latest aspects of IT Service Management at CMG's Annual Conference - www.cmg.org/conference

and validation would certainly be desirable, with enhance forecasts of new-user impact, both in terms of
interfa~es to other modeling techniques (e.g., queueing resources and costs. At the very leaat, it supplies an
theory) 8 logical possibility. The structure of the organized, documented insight into the nature of the
capacity-costinl system Areatly facilitates this type user baae.
of verification. Horeover, the overview nature of the

Buy the Latest Conference Proceedings and Find Latest Computer Performance Management 'How To' for All Platforms at www.cmg.org
system allows it to serve as a centralized tool for Theae enhancements, as well 8S the basic methodology,
typ1fying a -rriad of underlying diagnostic criteria can also be applied to the VH environment.
which 1D8Y be supporting it. Undoubtedly. special consideration ~st be liven to
guest operating .yste1ll8 (DOS. VS1. etc.) and other
One of the W»re proldsing applications of a fully aspects of VM operation which .ight distln,uish it from
verified and accepted capacity-eoeting system is shown typical HVS activity. In any case, the ba.ic
in Figure 11. Here the basic lDlltrix is reformatted philosophy, concepts, and techniques of
tnto a higher-level su.mary, arranged as a user capaCity-costing form a disciplined. productive method
profile. This can provide highly productive, of analyzing VH. Here too, as with KVS, this .ethod
Join over 14,000 peers - subscribe to free CMG publication, MeasureIT(tm), at www.cmg.org/subscribe

quick-reference visibility to finance, technology, provides a common frame of reference for technology and
una,eMnt, and individual users for a varlety of finance in reaching sound, coordinated management
purposes. Catalop of such profiles would greaUy deci8ions,

USER PROfilE

usor ~ 105/82) X
User Mew User _)'815 _1'10
Mntch~up: _

x
+~~-- .. _---_.-~-+
ISERVICE SUMMARY 1
Resource:
I lOll-UP)
_eru ti.e
_tutu' [XCI's
+-~ •• ~ ••_•• ---.-+ _roa' IIltIIOf'Y

I BATell 1 IHl£RACTlVE I TRAHSACTlotl

--~~~-!.--l
(U800·100o)l
-I
1'
i--------
I
I 30.5037 rCR soc. 1l.lZ5 leD su.1 63,211 leo sec.
, I
~
,
.._. __ ....+--.._----.. I
~--~._--+~--_._.-_.~. __ ._--+_...__._-_..
I
_-~---.

--~~~~-~---t
t 1800-2"00) I
II

I 10.6111 ICD sec. 2.302 TeO sec'l 2~.9"6 TCB sec.

I I I
-----------+--~-~------~-------+~.---~_
Zone 1 1
.. _-~ +.~._ -._ ---
-----------1
121100-0800) I
,II III
1 1l.190 T~8 sec. I 2 .....9 Teo soe. 1 22._508 TeB soc.
I I I
I I 1
-----------+-------------------+------------------+-------------------
lono II I 1 I
-----------1 I I
lwato;endsl 1 I 1
I 45,551 TeB soc. I 2.316 TCB sec. I 80,9'9 TCD 5ec.
I I ,
_ _ _ _ 1I II I
1 _

»> MAJOR APPLICATIONS; CICS. IOMS. [ASVTR[V. fIL[DIT

+-----------------+
I UAS( COS I SUl1HAltv 1
.-.. ~-----_.-~----+

01' Zom:: By Stlr\licu:

1'0110 1 611;
IG I
S ?ll. "~<J Dalcl, " 23 S s 12,120

,n ~ s '.9.n?"
lon(J 2
lone 3
ZOIliO ..
20, ?')]
l#I.?(,U
InLnrncl.
I
::: '5 S
rllnSlict ::: 62 I i 'lei,56]
196,613

lOU" lOU"

AI.Uchttd are: capacIL)'-cUSlillg pr-of'le

cftpllclLy 'MpuCt profile
I IIcr_IlIIL3 I I;u~t 5u_ry 1_5 .ppruprluu,l

PIG U R Z 11

296

Find a CMG regional meeting near you at www.cmg.org/regions