cloud computing, soa and web 2.0, an inevitable convergence
DESCRIPTION
This presentation talks about software and hardware design oriented towards embracing the cloud. The main point is that a service oriented architecture is the chief enabler to leverage cloud technologies on both the software and the hardware levels.TRANSCRIPT
D A V I D K A R A M
Cloud Computing, SOA and Web 2.0An Inevitable Convergence
W H Y S H O U L D I T H I N K O F M Y C O M P U T A T I O N A L S O F T W A R E A S A C O M P U T A T I O N A L S E R V I C E ?
H O W C A N A N D W H E N S H O U L D I O U T S O U R C E C O M P U T A T I O N S T O T H E C L O U D ?
H O W C A N I L E V E R A G E W E B 2 . 0 T O B R I D G E T H E G A P B E T W E E N A L G O R I T H M S A N D S C I E N T I F I C U S E R S ?
Answers We Were Aiming For
Neural Networks – Quick Intro
Algorithms for artificial intelligence
Typical applications
System identification– e.g. Stock prediction
Control – e.g. Robotics
Data mining – e.g. Knowledge discovery in databases
Exposing Kernels as Services
W H Y S H O U L D I T H I N K O F M Y C O M P U T A T I O N A L S O F T W A R E A S A C O M P U T A T I O N A L S E R V I C E ?
H O W C A N A N D W H E N S H O U L D I O U T S O U R C E C O M P U T A T I O N S T O T H E C L O U D ?
H O W C A N I L E V E R A G E W E B 2 . 0 T O B R I D G E T H E G A P B E T W E E N A L G O R I T H M S A N D S C I E N T I F I C U S E R S ?
The Basic Problem – Distant Kernel
Kernel
The Basic Problem – Distant Kernel
Kernel Application
The Basic Problem – Distant Kernel
Kernel
The Basic Problem – Distant Kernel
Kernel Application
The Basic Problem – Distant Kernel
Inherent mismatch with outside world
Mismatched languages
Mismatched hardware
Total lack of Entry Points / API’s
Java MLC++NN
Level 1: Exposing the Kernel
Java MLC++ NN
Clean and robust API
RPC bedStandardized Web Service
System View
C++ NN
NN
Level 1
Exposing the Kernel
Standardized Web Service
Lesson LearnedService orientation grants modularity to leverage right tools
Browser
GUI
Java
Higher level AI
C++
Number Crunchers MATLAB
Prototype
R
Statistical Module
Use the right platform and the right tools for the right job!
Broker standardized resources into a homogeneous whole!
Where can this be useful?
Where can this be useful?
Where can this be useful?
Where can this be useful?
Managing Computations in the Cloud
W H Y S H O U L D I T H I N K O F M Y C O M P U T A T I O N A L S O F T W A R E A S A C O M P U T A T I O N A L S E R V I C E ?
H O W C A N A N D W H E N S H O U L D I O U T S O U R C E C O M P U T A T I O N S T O T H E C L O U D ?
H O W C A N I L E V E R A G E W E B 2 . 0 T O B R I D G E T H E G A P B E T W E E N A L G O R I T H M S A N D S C I E N T I F I C U S E R S ?
The Basic Problem – Amassing Resources
NN Local
The Basic Problem – Amassing Resources
NNLocal
Grid
The Basic Problem – Amassing Resources
NN Local
Grid
Cloud
The Basic Problem – Amassing Resources
NN Local
Grid
Cloud
Level 2: Distributing the Computations
Outsource computations when local resources get burdened
TUM
Public Cloud
InternetEnsemble Run Client
Siemens
System View
Level 1
Exposing the Kernel
Level 2
Integrating Distributed Applications
Siemens
TUM
Public Cloud
InternetEnsemble Run Client
C++ NN
NN
WS
Lesson LearnedA SOA is the natural fit for leveraging cloud resources
Browser
GUI
Local
Computational Cores
Cloud
Extra Power Notch
Cluster
Data Mining
GPU
Linear Algebra
Use the right hardware for the right task!
Science and the Web Experience
W H Y S H O U L D I T H I N K O F M Y C O M P U T A T I O N A L S O F T W A R E A S A C O M P U T A T I O N A L S E R V I C E ?
H O W C A N A N D W H E N S H O U L D I O U T S O U R C E C O M P U T A T I O N S T O T H E C L O U D ?
H O W C A N I L E V E R A G E W E B 2 . 0 T O B R I D G E T H E G A P B E T W E E N A L G O R I T H M S A N D S C I E N T I F I C U S E R S ?
System View
Level 1
Exposing the Kernel
Level 2
Integrating Distributed Applications
Level 3
Web Integration
SENN
Siemens
TUM
Public Cloud
Internet
Ensemble Run
Client
WEB
User
Final Notes
Coupling SOA, Cloud & Web 2.0
Use established web standards to export and import computational algorithms in the cloud
Achieve a new dimension of modularity for software and hardware requirements
Use the intuitiveness of the web to bridge the gap between algorithms and scientists
move science into the web browser!
Questions