Dare il benvenuto i Soci di Sherwood

1
(No Transcript)
2
Dare il benvenuto i Soci di Sherwood la Tournée
di Studio 53 Ospiti italiani

• La Pianificazione di capacità e l'E-COMMERCIO
  Dirigere il Negozio Attraverso una Conoscenza
  Infrastructure Basato
• è Entrato,
• la Chiamata In,
• il Ceppo Su

3
Welcome Sherwood Associates Study Tour
53Italian Guests

• Capacity Planning and E-Commerce
• Managing the Store Through a
• Knowledge Based Infrastructure
• Come In,
• Call In,
• Log On

4
Agenda Purpose

• Brief Review of Capacity Planning Methodology and
  Objectives
• Statistical Process Control SPC what is it and
  how it provides a value add
• JCPenney E-Business
• BMC Patrol, SAS SPC demo

5
L'agenda lo Scopo

• La Revisione breve di Methodology di
  Pianificazione di Capacità e
• gli Obbiettivi il Controllo di Processo
  Statistico SPC ciò che è
• esso e come fornisce un valore aggiunge
  la Pattuglia di BMC DI
• E-AFFARI DI JCPenney, SAS il demo di SPC

6
Capacity Planning Methodology

• In existence since early 80s.
• Forecast based on expected business requirements.
• Focus on resource consumption.
• Full support by senior management, a mandate.
• Tool set BMC/Best1/Patrol/KMs, BEZ, Strobe, PAIO
  driver, SAS/SPC,

7
Methodology di Pianificazione di capacità

• Nell'esistenza da presto 80s. La previsione ha
  basato sui requisiti
• di affari aspettati. Mettere a fuoco sul consumo
  di risorsa. Il
• sostegno pieno dalla gestione superiore, un
  mandato. La serie di
• attrezzo BMC/Best1/Patrol/KMs, BEZ, Strobe,
  l'autista di PAIO,
• IL SAS/SPC,

8
SENIOR MANAGEMENT EXPECTATIONS OF CAPACITY
PLANNING

• Produce annual 18 month forecast!
• Used as input to ITB and acquisition process.
• Validate and review forecast against actual usage
  on monthly basis.
• Produce daily SPC based capacity updates for all
  E-Business sites.
• Also, REAL TIME MONITORING ALERTING OF CRITICAL
  SERVICE ASSURANCE METRICS
• Prove the forecast wrong!
• Ensure efficient and effective usage of all
  computing resources!

9
LAS ESPERANZAS MAYORES DE LA ADMINISTRACION
DE LA PLANIFICACION DE LA CAPACIDAD

• PRODURRE ANNUALE 18 PREVISIONE DI VALIDATE DI
• PREVISIONE DI MESE ALL'USO REALE SPC DI
• CAPACITÀ MENSILE QUOTIDIANO HA BASATO GLI
• AGGIORNAMENTI PER GLI E-AFFARI CONTROLLO
  DI
• TEMPO REALE METTERE ALL'ERTA DI
• ASSICURAZIONE DI SERVIZIO CRITICA METRICA
• PROVA L'INGIUSTIZIA DI PREVISIONE
  ASSICURA
• L'USO EFFICIENTE ED EFFICACE DI TUTTE LE
  RISORSE
• che CALCOLANO

10
JCPenney Computing Infrastructure

• Four major processing locations
• ITCC central command center 8,760 hr/yr.
• 4,500 Mainframe MIPS, 30 TBs storage, 21 STK
  Silos
• Major SNA / SNI Network
• 17 VTAM Images, 29 NCPs, 23,951 SNA Res.
• TCP/IP LAN/WAN Network
• 3,951 Routers, 69,135 Hubs Switches
• Multiple T3 T1 Circuits

11
JCPenney Computing Infrastructure cont.

• 9 AS400s - Specialty Retailing / Prescription
  Adjudication
• Over 3,000 NT, 200 Unix servers.
• People Soft, Intranet, POS, Esbase, Oracle,
  SQLserver, XML, 100k email users, Cognos
• 20 Node Teradata - MPP Data Mining
• 48 Tandems - Credit Authorization
• 6 Ecommerce sites by end of year 1999

12
(No Transcript)
13
Statistical Process Control (SPC)What is it?

• A branch of mathematics developed by Dr. Shewhart
  over 60 yrs ago.
• Used by Dr. Demming in Japan to improve
  manufacturing after WWII.
• Dr. Wheeler is current biggest proponent today.
• WWW.SPCPRESS.COM

14
Il Controllo di Processo statistico (SPC) ciò che
è esso?

• Un ramo di mathematics sviluppato da Dott.
  Shewhart sopra 60
• yrs fa. Usato da Dott. Demming in
  giappone di migliorare la
• fabbricazione dopo WWII. Dott. Wheeler è
  proponente attuale il
• più grande oggi. WWW. SPCPRESS. COM

15
(No Transcript)
16
Statement of what SPC is for Capacity Planning

• SPC is a methodology used to determine when a
  chronological set of measurements from a process
  are outside the bounds of variation inherent to
  the process. A measurement that exceeds the
  bounds is said to be out of control. SPC has been
  traditionally applied to manufacturing processes.
  However, we have not seen any convincing evidence
  that it will not work in an information systems
  environment. If you view the task of information
  systems as manufacturing information from raw
  data, it is not a very long leap. -Mich
  ael Gresham, Senior Capacity Planner JCPenney

17
L'istruzione di che SPC è per la Pianificazione
di Capacità

• SPC è un methodology ha usato per determinare
  quando una
• serie cronologica di misure da un
  processo sono fuori dei limiti di
• variazione inerente al processo. Una
  misura che eccede i limiti è
• detta per essere fuori di controllo. SPC
  è stato tradizionalmente
• applicato alla fabbricazione di processi.
  Comunque, non abbiamo
• visto qualunque evidenza convincente che
  non lavorerà in un
• ambiente di sistemi di informazioni. Se
  lei osserva l'incarico di
• sistemi di informazioni informazioni come
  fabbricando dai dati
• crudi, non è un salto. molto lungo
  -Michael Gresham, la Capacità
• Planner JCPenney Superiore

18
What we get from SPC!

• TURN INFORMATION INTO KNOWLEDGE
• ENSURE EFFICIENT USE OF RESOURCES
• QUANTIFY THE COSTS OF COMPUTING
• DETERMINE PREDICTABILITY vs. UNPREDICTABILITY
• FACILITATE CONTINUAL PROCESS IMPROVEMENT
• SAVE MONEY

19
Case StudyUsing SPC in conjunction with
traditional tools for extracting information from
performance metric data
20
Il caso StudyUsing SPC nella congiunzione con gli
attrezzi tradizionali per estrarre di
informazioni dall'esecuzione dati metrici
21
POSI Extended Batch Run TimesProblem Scope

• A new system went live in 4Q98 with full rollout
  achieved June 99.
• Concerns raised that Batch stream ran too long
  compromising SLU of Daily Sales postings to WEB.
• No correlation between CPU Busy and elapsed time
  per qualifier.
• Multiple Strobe runs indicate the majority of the
  wait time is in UDS.
• The problem is most acute in Lenexa.
• Capacity Planning began detailed analysis late
  June 99.

22
Analysis Process

• Average Batch elapsed 1.5 to 2 hours
• Less that 60 seconds of cpu consumed per job.
• I/0 subsystem and memory eliminated as
  degradation contributors.
• Strobe runs determined a user written SVC and
  subsystem called UDS was primary cause of wait
  time.

23
Summary of Changes for POSI Job Stream prior to
GRS Star

• Technical Support changed the UDS allocation
  parameters 07/27/1999
• Reduced number of catalogued UDS elements filling
  up user catalogs
• Reorganized user catalogs
• No improvement in job elapsed time.
• POSI developers rearranged the POSI job activity
  in July
• Allows New POSI to run sooner.
• Window starts earlier but no improvement in job
  elapsed time.
• Dispatch priority changed for POSI job stream
  07/28/1999
• Ensured POSI dispatching higher priority than
  normal batch
• No improvement in job elapsed time.
• Technical Support increases the priority of UDS
  08/12/1999
• No improvement in job elapsed time.
• Technical Support changed a UDS wait time value
  08/13/1999
• Value reduced from 5 seconds to 1 second.
• Minimal improvement in job elapsed time.

24
(No Transcript)
25
GRS RESMIL change 10 mils to 1 mil Percent
Improvement 81
26
(No Transcript)
27
GRS Ring to Star Percent improvement 71
28
(No Transcript)
29
GRS RESMIL change 10 mils to 1 mil Percent
improvement 23
30
(No Transcript)
31
(No Transcript)
32
Nominal Change in CPU consumed
33
(No Transcript)
34
Ring to Star GRS change Percent Improvement 15
35
(No Transcript)
36
(No Transcript)
37
Nominal Change in CPU consumed
38
DALLAS GRS Ring to Star October 17 Metric is
INIT-DS Workload is DMS TSO jobs
39
Conclusions

• Elapsed time per qualifier in Lenexa improved by
  83 .
• POSI jobs in all locations now complete on
  average in 21 minutes.
• Overall all Batch benefited
• Elapsed time reduced 23
• UDS intensive job throughput increased to over
  1000 per hour
• CPU usage change was nominal.
• SPC provides quantitative results

40
Le conclusioni

• Il tempo di Elapsed per il qualifier in Lenexa ha
  migliorato da 83
• . I lavori di POSI in tutte le posizioni
  completano adesso sulla
• media in 21 minuti. Globale tutto il
  Lotto giovato il tempo di
• Elapsed ha ridotto 23 UDS il
  throughput di lavoro intenso
• aumentato a sopra 1000 per il cambiamento
  di uso di CPU di ora
• era nominale. SPC fornisce dei risultati
  quantitativi

41
POSI Metrics Recap with Senior
ManagementThe Dog Is Dead Rick Conley
Director of Systems and Architectures JCPenney,
Information Systems

• Capacity Planning Dept.

42
WWW.JCPENNEY.COM
The Look, Look Who
Come In Call In Log On
43
JCPenney.com E-businessStrategic Goals

• Providing access to the entire JCPenney
  assortment (catalog and store)
  
• Driving sales by highlighting areas of
  JCPenney.com distinctiveness
• Creating an intuitive, easy shopping experience
• Extending product lines/site offerings beyond
  current assortment in areas of JCPenney.com
  distinctiveness
• Iterating merchandising focus based on monitoring
  and customer feedback
• Aggressively piggyback on catalog assortment/
  merchandising process

44
E-BUSINESS WWW.JCPENNEY.COM

• JCPenneys e-commerce technology stack
• Intel-based multiprocessor servers
• Microsoft Windows NT EE operating system
• Internet Information Server
• Active Server Pages for providing dynamic content
• Site Server for large-scale content distribution
• SQLServer Enterprise Edition
• Database replication for fail-over
• Live Picture product imaging
• NetPerceptions shopper intelligence
• HydraWeb LAN and WAN load balancing and traffic
  management
• BMC Patrol service monitoring and management

45
(No Transcript)
46
iStore Management Architecture
47
Four Tenets for Real-Time Service Management are
to Diagnose and Fix the Root Cause of Technology
Problems within the JCPenney Internet Store
1. Monitor the Internet Store and components
that comprise it.
2. Locate the lowest technology level
component contributing to the problem.
3. Document the process used by the support
staff to fix the problem.
4. Automate the fix so the that next outage or
degradation is minimized.
48
Live Patrol Demo WWW.JCPenney.com Internet
Store total sales is affected by...
49
SPC - The Key to Determining Whats Normal and
Whats Not

• Statistical Process Control (SPC) is the
  mechanism used by the Internet Store management
  team to determine predictable vs. non-predictable
  conditions, plan for future capacity, turn
  information into knowledge and gain a competitive
  advantage.
• WEB/SPC Internet Store web page
• SPC results can be dynamically inserted into the
  real-time management component for intelligent
  notification vs. static thresholds for a
  constantly changing environment.

50
A Winning Combination
The combination of real-time management,
application service automation, distributed
notification and statistical process control
provide an extremely comprehensive management
solution for electronic business environments.
51

• SPC helps JCPenney cut costs and
  boost performance.
• -Dave Evans
• JCPenney Senior Vice President and CIO

52
SPC aiuta l'esecuzione. di costi di taglio di
JCPenney ed incremento

• -Il Dave Evans
• JCPenney il Presidente di Vizio Superiori e CIO

53
A single accurate measurement is worth a
thousand expert opinions. -Grace Hopper
(1906-1992)
54
Una misura sola esatta vale mille opinioni.
esperto -Hopper di Grazia (il 1906-1992)
55
Identifying the characteristics and causes of
change is the key to the management of complex
systems. SPC provides that key! -Dr. H. Pat
Artis CEO, Performance Associates, Inc.
56
Identificare le caratteristiche e le cause di
cambiamento è la chiave alla gestione di sistemi
complicati. SPC fornisce quella chiave!
-Dott. H. Carezzare Artis CEO, i Soci di
Esecuzione, Inc.
57
Conclusion

• By exploiting SPC we are able to separate Black
  and White from shades of Gray.
• Turn Information into Knowledge.
• Through a combination of BMCs Patrol KMs and SPC
  we are gaining a Competitive Advantage to Serve
  Our Customers Better.