13
Wolfgang Gentzsch & BurakYenier The UberCloud Inc. The UberCloud 2012 2015 From Experiment to Marketplace Our ‘last mile’: application containerization Recorded for the RichReport, January 2015

UberCloud: From Experiment to Marketplace

Embed Size (px)

Citation preview

Page 1: UberCloud: From Experiment to Marketplace

Wolfgang  Gentzsch    &    Burak  Yenier  The  UberCloud  Inc.  

The  UberCloud    2012  -­‐  2015  

From    Experiment    to    Marketplace  

Our  ‘last  mile’:  application  containerization  

Recorded    for  the    RichReport,  January  2015  

Page 2: UberCloud: From Experiment to Marketplace

Introduction  

Ê  The  UberCloud  enables  companies  to  build,  deploy,  and  consume  compute  and  data  intensive  applications  as  a  service:    better,  faster,  and  cheaper    

Ê  Experiment  started  July  2012  to  learn  HPC  Cloud  challenges:  now  166  experiments,  54  cloud  providers,  80+  ISVs,  100s  of  experts,  2500  participants  

Ê  Computing  scenario:  workstations,  servers,  and  clouds  

Ê  Pain:  workstations  limited,  servers  expensive,  clouds  new  

Ê  Solution:  Application  containerization  for  UberCloud  Marketplace    

Page 3: UberCloud: From Experiment to Marketplace

Sponsors  

 

 

 

,  partners  ,  customers  

Page 4: UberCloud: From Experiment to Marketplace

 HPC  in  the  Cloud:    benefits  

Continue  using  your  workstation  (and  your  server)  for  your  daily  design,  and  use  Cloud  resources  with  additional  benefits:  

Ê  An  HPC  system  at  your  finger  tip,  on  demand,  as  you  need  it  

Ê  Pay  per  use:  no  CAPital  EXpenditure  

Ê  Business  flexibility:  Scaling  resources  up  and  down    

Ê  Low  risk:  working  with  multiple  cloud  providers  

Ê  Latest  technology:  cloud  providers  upgrade  continuously  

Page 5: UberCloud: From Experiment to Marketplace

HPC  in  the  Cloud:    challenges  

Ê  It’s  a  new  business  and  working  paradigm  

Ê  Cloud  expertise  needed  

Ê  Access  to  clouds  is  cumbersome    

Ê Which  cloud  is  right  for  me?  

Ê  Loosing  control  over  your  jobs  and  data  

Ê  Security,  privacy,  trust  in  service  provider  

Ê  Intellectual  property  

Ê  Software  Licensing  

Ê Heavy  data  transfers  

 

Page 6: UberCloud: From Experiment to Marketplace

Solution:  UberCloud  SaaS  Factory  

UberCloud        SaaS  Factory  

Software   SaaS  

Commercial  ISV  Open  Source  In-­‐house  

Public  Cloud  Private  Cloud  On-­‐Premise  

$$  $  

We  turn  high-­‐TCO,  complex  software  into  user-­‐friendly  pay-­‐per-­‐use  services:  

Development  Services  (Docker)  

-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  Deployment  &  

Management  Services  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  Marketplace  

   

Providers                                    Consumers  

$  

   UberCloud  SaaS  Factory  

Page 7: UberCloud: From Experiment to Marketplace

Builder  

Launcher  

Controller    ISV   Data  Tools  

Stackable  units  with  tools  (ex:  encryp?on),  ISV  applica?on  codes    (ex:  OpenFOAM).  Just  add  your  own  codes  and  data.    

Run  anywhere  with  UberCloud  Run  Time.  Scale  up  or  down  the  compute  power  as  needed.  

Collect  granular  usage  data,  logs.  Monitor,  alert,  report.    

On-­‐Premise   Private  Cloud  

Public  Cloud  

Run  Time   Run  Time   Run  Time  

Removing  the  Bottlenecks  with  Containers:  Build  once,  run  anywhere  

Page 8: UberCloud: From Experiment to Marketplace

Builder  

Launcher  

Controller    ISV   Data  Tools  

Stackable  units  with  tools  (ex:  encryp?on),  ISV  applica?on  codes    (ex:  OpenFOAM).  Just  add  your  own  codes  and  data.    

Run  anywhere  with  UberCloud  Run  Time.  Scale  up  or  down  the  compute  power  as  needed.  

Collect  granular  usage  data,  logs.  Monitor,  alert,  report.    

On-­‐Premise   Private  Cloud  

Public  Cloud  

Run  Time   Run  Time   Run  Time  

 Containers:  Build  once,  run  anywhere  

Page 9: UberCloud: From Experiment to Marketplace

Builder  

Launcher  

Controller    ISV   Data  Tools  

Stackable  units  with  tools  (ex:  encryp?on),  ISV  applica?on  codes    (ex:  OpenFOAM).  Just  add  your  own  codes  and  data.    

Run  anywhere  with  UberCloud  Run  Time.  Scale  up  or  down  the  compute  power  as  needed.  

Collect  granular  usage  data,  logs.  Monitor,  alert,  report.    

On-­‐Premise   Private  Cloud  

Public  Cloud  

Run  Time   Run  Time   Run  Time  

 Containers:  Build  once,  run  anywhere  

Page 10: UberCloud: From Experiment to Marketplace

Builder  

Launcher  

Controller    ISV   Data  Tools  

Stackable  units  with  tools  (ex:  encryp?on),  ISV  applica?on  codes    (ex:  OpenFOAM).  Just  add  your  own  codes  and  data.    

Run  anywhere  with  UberCloud  Run  Time.  Scale  up  or  down  the  compute  power  as  needed.  

Collect  granular  usage  data,  logs.  Monitor,  alert,  report.    

On-­‐Premise   Private  Cloud  

Public  Cloud  

Run  Time   Run  Time   Run  Time  

 Containers:  Build  once,  run  anywhere  

Page 11: UberCloud: From Experiment to Marketplace

UberCloud  Containers:    Reducing  /  Removing  Cloud  Challenges  

http://www.theubercloud.com/workstations-­‐servers-­‐clouds-­‐comparing-­‐apples-­‐apples/    

Page 12: UberCloud: From Experiment to Marketplace

UberCloud  Marketplace  

Page 13: UberCloud: From Experiment to Marketplace

ANSYS  Container  Demo  

Ê  ANSYS  multi-­‐physics,  Mechanical  ,  Fluent,  CFX,  Workbench,  Icepak,  EM,  LS-­‐DYNA  and  LS-­‐PrePost    

Ê  All  running  inside  one  Docker  container,  in  the  cloud  

Ê  Pre-­‐  and  post-­‐processing  in  the  cloud  work  the  same  way  that  it  would  work  on  the  engineer’s  desktop  

Ê User  experience  in  the  cloud  matches  that  on  the  desktop  

Ê GPU  support  for  high-­‐resolution,  fast,  remote  visualization  

Ê  Automated  monitoring  keeps  you  connected,  under  control  Ê DEMO:  http://youtu.be/-­‐5HV69Bjm20        (4  minutes)