financial,(telco,(retail,(&( manufacturing:(hadoop...

25
1 © Cloudera, Inc. All rights reserved. Financial, Telco, Retail, & Manufacturing: Hadoop Business Services for Industries Ho Wing Leong, ASEAN

Upload: others

Post on 27-Jun-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

1  ©  Cloudera,  Inc.  All  rights  reserved.  

Financial,  Telco,  Retail,  &  Manufacturing:  Hadoop  Business  Services  for  Industries  Ho  Wing  Leong,  ASEAN    

Page 2: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

2  ©  Cloudera,  Inc.  All  rights  reserved.  

Cloudera  company  snapshot  

Founded  2008,  by  former  employees  of  Company  Largest  Hadoop  Company  Globally  Employees  Today  800+  worldwide  World  Class  Support  More  than  100  24x7  global  staff  

Pro-­‐acQve  &  predicQve  support  programs  using  our  EDH  Mission  CriQcal  ProducQon  deployments  in  run-­‐the-­‐business  applicaQons  

worldwide  –  Financial  Services,  Retail,  Telecom,  Media,  Health  Care,  Energy,  Government  

The  Largest  Ecosystem  More  than  1,450  Partners  Cloudera  University  Over  40,000  trained  Open  Source  Leaders  Cloudera  employees  are  leading  developers  &  contributors  to  

the  complete  Apache  Hadoop  ecosystem  of  projects.  

Page 3: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

3  ©  Cloudera,  Inc.  All  rights  reserved.  

Customer  success  across  industries  

The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it

Financial  Services  

Telecom  

Healthcare  &  Life  Sciences  

Media  &  Technology  

Retail  &    CP  

Public    Sector  

Page 4: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

4  ©  Cloudera,  Inc.  All  rights  reserved.  

Explore  the  PossibiliQes  of    SAS  and  Cloudera  • The  combinaQon  of  SAS  analyQcs  and  Cloudera’s  Enterprise  Data  Hub  (EDH)  is  a  common  recipe  for  AnalyQcs  at  Scale.    • While  Cloudera’s  EDH  makes  it  feasible  and  economically  viable  to  store  and  manage  extreme  volumes  of  data  in  one  place,  SAS’  In-­‐Memory  AnalyQcs  gives  you  the  power  to  analyze  and  mine  data  at  Scale  …  all  on  a  single  system.    

Page 5: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

5  ©  Cloudera,  Inc.  All  rights  reserved.  

SAS  &  Cloudera  Partnership  The  Tightest  Product  Level  IntegraQon  

Execu8ve  sponsored  partnership  which  spans  R&D,  Product  Management,  Sales,  Marke8ng,  Consul8ng  &  Educa8on  Services.    SAS  product  integra8on  with  Cloudera  is  the  most  extensive  of  all  the  commercial  Hadoop  distribu8ons    •  SAS  internal  development  teams  have  a  Cloudera  first  policy  and  all  internal  work  is  performed  on  Cloudera  clusters.  •  Dedicated  Cloudera  resources  at  Cloudera  HQ  and  SAS  HQ  working  with  SAS  R&D  •  SAS  has  dedicated  R&D  resources  to  opQmize  SAS  soluQons  for  the  Cloudera  pladorm  •  Pordolio  includes  integraQon  with  Access  to  Hadoop,  Access  to  Cloudera,  Visual  AnalyQcs,  In-­‐Memory  StaQsQcs,  High  Performance  AnalyQcs,  Scoring  Accelerator  for  Cloudera  Hadoop  &  Visual  StaQsQcs  among  others…  

Page 6: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

6  ©  Cloudera,  Inc.  All  rights  reserved.  

SAS  &  Cloudera  Partnership  Strong  Go  To  Market  Alignment  

•  Engineering  schedule  coordinaQon  to  ensure  quick  uptake  of  new  releases  from  each  side  •  SAS  /  Cloudera  Webinar  Series    •  Reciprocal  Services  Agreement  in  place    •  Joint  Training  course  developed  to  provide  educaQon  on  Cloudera  Hadoop  and  SAS  content  for  analyQcs  on  big  data  •  SAS  SoluQons  OnDemand  Preferred  Vendor  is  Cloudera  •  SAS  Visual  AnalyQcs  and  Cloudera  Enterprise  Data  Hub  Starter  Service  package  

Cloudera  and  SAS  ConfidenQal  

Page 7: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

7  ©  Cloudera,  Inc.  All  rights  reserved.  

SAS  &  Cloudera  SoluQon  Stack  

Next-­‐Genera8on  SAS

®  User  

SAS®  User  

MPI  Based  

User    Interface  

Metadata  

Data    Access  

Data  Processing  

File  System  

SAS®  LASR™  AnalyQc  Server  

HDFS  

Base  SAS  &  SAS/ACCESS®  Interface  to  Hadoop™  

SAS  Metadata  

Pig  

Map  Reduce  

In-­‐Memory  Data  Access  

SAS®  Display  Manager   SAS®  Visual  AnalyQcs  SAS®  Enterprise  Miner™  

SAS®  Data  IntegraQon  

SAS®  Enterprise  Guide®  

Hive  SAS  Embedded    

Process    DS2  Accelerators    

 

SAS®  High-­‐  Performance  

AnalyQc    Procedures        

HBASE   Impala    

Page 8: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

8  ©  Cloudera,  Inc.  All  rights  reserved.  

Three  Factors  Entrenching  Big  Data  in  Financial  Services  1.  Compliance  and  Strategy:  Growth  in  a  Stringent  Regulatory  Environment  

Accenture  and  CEB  TowerGroup  say…  

Sources:  Dash,  Eric.  “FeasQng  on  Paperwork,”  The  New  York  Times.  September  8,  2011.                                  Accenture.  Coming  to  Terms  with  Dodd-­‐Frank.  January  2013.    

believe  Dodd-­‐Frank  will  strengthen  their  compeQQve  posiQoning  

 agree  Dodd-­‐Frank  will  benefit  their  own  company’s  customers  

 anQcipate  spending  $50  million  or  more  on  compliance  

64%    

83%    

50%  

Page 9: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

9  ©  Cloudera,  Inc.  All  rights  reserved.  

Three  Factors  Entrenching  Big  Data  in  Financial  Services  2.  Mass  PersonalizaQon:  Tailoring  Products  and  Services  Across  the  Value  Chain  

DeloiVe  and  Core  Profit  say…  

Average  customer  acquisiQon  costs  retail  banks  

MORE  THAN  $350  and  requires  customers  to  

carry  balances  NEARING  $10,000  just  to  break  even  

 Sources:  Deloire.  2014  Banking  Industry  Outlook.  February  2014.                                  Andera  &  CoreProfit.  The  Future  of  Account  Opening  2011.  June  2011.    

Page 10: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

10  ©  Cloudera,  Inc.  All  rights  reserved.  

Three  Factors  Entrenching  Big  Data  in  Financial  Services  3.  Towards  CompeQQve  Advantage:  ConsolidaQon  Around  High-­‐Return  OpportuniQes  

Morgan  Stanley  Research  and  Oliver  Wyman  say…  

During  the  past  20  years,  the  margins  on  deposits  and  cash  

equiQes  have    DECLINED  BY  33%  TO  50%  while  the  need  for  compuQng  

power  in  FinServ  has    GROWN  200%  TO  500%  FASTER  THAN  REVENUE  

Sources:  Morgan  Stanley  Research  &  Oliver  Wyman.  Wholesale  and  Investment  Banking  Outlook  2014.  March  2014.                                    Oliver  Wyman.  The  State  of  the  Financial  Services  Industry  2013.  January  2013.  

Page 11: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

11  ©  Cloudera,  Inc.  All  rights  reserved.  

Customer  Experience  Mgmt.  (Customer  360)   Network  OpQmizaQon  

Data  MoneQzaQon  OperaQonal  AnalyQcs  

Pusng  Big-­‐Data  to  Work  for  Telcos  Key  Use  Cases  and  Areas  of  ApplicaQon  for  Today’s  Telcos  

Page 12: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

12  ©  Cloudera,  Inc.  All  rights  reserved.  

Customer  Experience  Mgmt.  (Customer  360)   Network  OpQmizaQon  

Data  MoneQzaQon  OperaQonal  AnalyQcs  

Customer  Churn  

AnalyQcs  ProacQve  Care  

Targeted  MarkeQng/  

PersonalizaQon  

Network  Investment  &  Planning  

Real  –Time  Network  AnalyQcs  

Capacity  Planning  &  OpQmizaQon  

Revenue  Leakage/  Assurance  

Enterprise  Security  AnalyQcs  

Order  Management  

Data  AnalyQcs  As  A  Service  (DAaaS  

Geo-­‐LocaQon  as  a  Service  

VerQcal  Services  

Pusng  Big-­‐Data  to  Work  for  Telcos  Key  Use  Cases  and  Areas  of  ApplicaQon  for  Today’s  Telcos  

Page 13: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

13  ©  Cloudera,  Inc.  All  rights  reserved.  

Data  Sources  

Data  Systems  

Data  Access  

Business  AnalyQcs  

Custom  ApplicaQons  

ExisQng  Data  

Databases  

OperaQonal  ApplicaQons  

New  Data  

Tradi&onal  Architectures  Under  Pressure

Limited  Data  Not  efficient  to  keep  exis&ng  data,  let  alone  handle  new  data  sources. Time  consuming  to  transform  data  for  analysis  in  exis&ng  systems.

Limited  Insights  Power  users  struggle  with  data.

Many  users  have  no  data.

Compliance  and  Privacy  More  data,  more  users,  and  more  tools  create  complexity. Need  to  balance  business  agility  with  security  and  governance.

Page 14: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

14  ©  Cloudera,  Inc.  All  rights  reserved.  

Data  Sources  

Data  Systems  

Data  Access  

Business  AnalyQcs  

Custom  ApplicaQons  

ExisQng  Data  

Databases  

OperaQonal  ApplicaQons  

New  Data  

More  Value  from  More  Data  for  More  Users,  in  Less  Time

Keep  Unlimited  Data  From  disparate  and  limited  views,

to  unlimited  informa&on  access.

Unlock  Value  from  Data  From  analy&cs  for  some,  to

insights  for  all.

Manage  Compliance  From  risk  due  to  regula&ons  and  customer  privacy  concerns,

to  trust  in  a  secure  and  compliant  plaGorm.

Enterprise  Data  Hub  

Security  and  AdministraQon  

Unlimited  Storage  

Process   Discover   Model   Serve  

Page 15: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

15  ©  Cloudera,  Inc.  All  rights  reserved.  

Data  Changes  How  We  Work  

Everything  that  can  be  measured  will  be  measured.    

Employees  and  customers  expect  more  personal  interacQons,  but  not  at  the  cost  of  their  privacy.    

The  most  innovaQve  companies  embrace  experimentaQon  and  agility.    

InstrumentaQon   ConsumerizaQon   ExperimentaQon  

Page 16: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

16  ©  Cloudera,  Inc.  All  rights  reserved.  

SFR  Telecom  

Customer  Spotlight  

Page 17: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

17  ©  Cloudera,  Inc.  All  rights  reserved.  

Create  shared  view  into  the  customer  journey  •  Must  collect  data  from  >1B  

events  generated  per  day    •  Shared  view  of  data  on  

products,  device  usage,  invoices,  contracts,  price  plans,  and  call  detail  records  

Cloudera  EDH  •  Real-­‐Qme,  self-­‐service  search,  

reporQng,  analysis  •  Secure  via  Sentry    

SoluQon  

Customer  Spotlight:  SFR  Telecom  

Improved  quality  of  support  &  network  ops  •  Berer  customer  experience  

Challenge   Benefit  

“Instead  of  upgrading  our  DW  environment  every  3  years,  the  system  will  deliver  opQmal  performance  for  8  or  9  years  now.”  

Page 18: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

18  ©  Cloudera,  Inc.  All  rights  reserved.  

Mastercard  

Customer  Spotlight  

Page 19: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

19  ©  Cloudera,  Inc.  All  rights  reserved.  

Joint  Customer  Spotlight:  MasterCard  

Fraud  costs  credit  card  issuers  approximately  

$10  billion  per  year  and  is  only  detected  at  a  

40%  rate.  

Most  detecQon  models  are  limited  by  the  

amount  of  data  that  is  available  for  analysis  at  

one  Qme,  which  is  constrained  by  extreme  

cost.  

Impala  extends  queries  to  data  sets  spanning  mulQple  years,  not  just  the  tradiQonal  weeks  and  months.  

SAS®  Visual  AnalyQcs  and  SAS  Visual  StaQsQcs.  SAS/ACCESS  

SoluQon  

Move  ETL  and  storage  jobs  to  Hadoop,  which  cuts  costs  and  Qmelines  significantly.  

More  data  is  held  in  acQve  archive,  both  in  original  

and  digested  formats,  so  it  is  available  for  future  analysis.  

Test  new  models  using  historic  data  on  an  ad  hoc  

basis  using  full,  live  data  sets  at  zero  marginal  cost  

Challenge   Benefit  

Test  new  models  using  historic  data  on  an  ad  hoc  basis  using  full,  live  data  sets  at  zero  marginal  cost  

Page 20: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

20  ©  Cloudera,  Inc.  All  rights  reserved.  

MarkeQng  

Problem  

SoluQon  

Partners  

Next  Best  Offer  Berer  profile  the  customer  and  use  collaboraQve  and  context-­‐based  filtering  to  offer  the  most  appropriate  product,  product  bundle,  or  offer  at  any  given  Qme.    

Too  Many  Sources  Disparate  data  is  hard  to  correlate  and  analyze  for  sufficiently  personalized  product  bundling,  cross-­‐sell,  and  up-­‐sell  opportuniQes  served  in  real  Qme.  

Stream  Processing  Spark  Streaming  is  used  to  calculate  pricing  occasions  in  real  Qme  based  on  live,  unstructured  data-­‐in-­‐moQon  from  the  web,  sensors,  mobile  devices,  etc.  

Use  Case  

Page 21: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

21  ©  Cloudera,  Inc.  All  rights  reserved.  

Supply  Chain  

Problem  

SoluQon  

Partners  

Event  CorrelaQon  to  Store  Traffic  Model  historical  store-­‐specific  sales  to  event  data  (e.g.,  weather,  disbursements,  TV)  to  opQmize  inventory,  assortments,  in-­‐store  merchandising,  and  staffing.  

Can’t  Scale  Beyond  Silos  Current  systems  can  not  integrate  social,  telemetric,  public,  and  log  data  in  real  Qme  with  historical  data  to  predict  sudden,  temporary  demand  shixs.  

Calculate  Anything  HBase  is  a  real-­‐Qme  database  accommodaQng  complex  historical  data.  Spark  and  Impala  converge  ETL,  analyQcs,  and  reporQng  for  on-­‐demand  modeling.  

Use  Case  

Page 22: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

22  ©  Cloudera,  Inc.  All  rights  reserved.  

AutomoQve  &  Industrial  

Problem  

SoluQon  

Partners  

ProacQve  Quality  Assurance  Build  machine  learning  algorithms  that  idenQfy  producQon  anomalies  prior  to  field  tesQng  and  find  performance  flaws  that  could  not  be  idenQfied  in  R&D.  

Silos  Limit  OpQons  Legacy  systems  hold  historical  data  from  producQon  line  telemetry,  factory  surveillance  and  sensors,  call  centers,  in-­‐car  telemaQcs,  etc.  That  data  is  useless  if  it  is  kept  offline  and  in  silos.  

Anomaly  DetecQon  Spark  includes  MLLib,  a  library  of  machine  learning  algorithms  for  large  data,  enabling  clustering  to  idenQfy  outliers  from  typical  producQon  parerns.  

Use  Case  

Page 23: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

23  ©  Cloudera,  Inc.  All  rights  reserved.  

The  Road  to  Success    

Reference  implementaQon  to  3  sources,  5  transforms,  1  target  Create,  execute,  test,  and  review  a  custom  ingesQon/ETL  plan  

Apply  SQL  to  much  larger  data  sets  with  Impala,  Hive,  and  Pig  Master  advanced  techniques  that  boost  Hadoop  accessibility  

Combine  batch  and  stream  processing  with  interacQve  analysis  OpQmize  applicaQons  for  speed,  ease  of  use,  and  sophisQcaQon  

DescripQve  AnalyQcs  Pilot  

Data  Analyst  Training  

Spark  Developer  Training  

Joint  SAS  &  Cloudera  Data  ScienQst  Training  Class  taught  on  SAS  tools  and  SAS  scripQng  language  running  in  the  Cloudera  Enterprise  Data  

Hub  

Joint  SAS  &  Cloudera  Visual  AnalyQcs  Starter  Package  will  allow  you  to  get  up  and  running  on  Visual  AnalyQcs    quickly  

SAS  &  Cloudera  Data  ScienQst  Class  

Visual  AnalyQcs  Starter  Bundle  

Page 24: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

24  ©  Cloudera,  Inc.  All  rights  reserved.  

Cloudera  has  trained  over  

40,000  people  on  Hadoop  since  

2009  

Big  Data  professionals  from  

60%  of  the  Fortune  100  have  arended  live  Cloudera  

training  

Industry  leading  training  and  university  program  

Source:    Fortune,  “Fortune  500  “  and  “Global  500,”  May  2012.  

Page 25: Financial,(Telco,(Retail,(&( Manufacturing:(Hadoop ...©"Cloudera,"Inc."All"rights"reserved." 1 Financial,(Telco,(Retail,(&(Manufacturing:(Hadoop(Business(Services(for(Industries"

Hadoop  is  at  the  heart  of  the  big  data  movement.    Nobody  knows  Hadoop  like  Cloudera.  Visit  the  Cloudera  booth  for  more  informaQon.  

[email protected]