Ct Cloud Computing.V1.0

Cloud Computing Application Cloud Computing Application in HK & Chinain HK & China

Francis NgFrancis NgCluster Technology LimitedCluster Technology Limited

AGENDA

Some History Why AMZN provides Public Cloud Service? Biz model used by IBM and AMZN Cluster Technology Limited

Background Client Descriptions Business Case: Bank of China

AGENDA

Some History

Problem with Grid Computing / Parallel Computing High admin cost

Difficult to use from lay users perspective

Users still have to be aware of the physical hardware layer - # of nodes, etc.

Cloud Computing Users are transparent to the physical hardware through

hardware virtualization

It scale-in / scale-out based on pre-defined usage criteria. Administration cost is decreased.

IT resources are better utilized.

Cloud Computing Structure

Virtualization

Physical Hardware

Workloads

Cloud services - Request Driven Provisioning

InnovationEnablement

SoftwareDevelopment

Web 2.0 DataIntensive

Processing

ScalableTransactionProcessing

VirtualStorage

Virtual Application

Server

Virtual Application

Server

Virtual Application

Server

VirtualNetworks

VirtualServers

MonitoringSLA

Capacity PlanningDynamic

Scheduling

SSrc: IBM Cloud Computingrc: IBM Cloud Computing 發中心發中心

Cost Structure of Public Cloud Service

• 2 Points to Note: Initial fixed cost is switched from users to service providersAs users become increasingly rely on server access, it is cheaper

for users to implement their own IT infrastructure

Cost Structure of Public Cloud andTraditional IT Expense

0

500

1000

1500

2000

2500

1 2 3 4 5 6 7 8 9 10 11No. of Users

IT E

xpen

se (

$)

Traditional Cloud

Fix

ed C

ost

Why AMZN provides Public Cloud Services?

From service providers’ viewpoints, it has high initial fixed cost, so why are they still doing it?

The players are: Google, Amazon, IBM They have already built a hugh IT infrastructure to handle their

peak load.

Internally, AMZN want to utilize their IT resources to save their IT expenditure

Cloud Service solves their own problem. And they could take advantage of their existing IT infrastructure by make business use of it

=> Public Cloud Services is born

Business Model on Cloud Services 1/2

Public Cloud Software as a Service (SaaS) – charge by app. usage/no. of

accounts Google App for Enterprises Saleforce.com

Platform as a Service (PaaS) – charge by partnership with system integrators, app. developed on the platform

Google App Engine Windows Azure Platform (include SQL services, sharePoint services,

Dynamics CRM) Amazon SimpleDB

Infrastructure as a Service (IaaS) – charge by usage (eg. Bandwidth, amount of data accessed/month)

Amazon S3, Amazon EC2

Business Model on Cloud Services 2/2

Private Cloud Offer solutions of hardware setup, software installation,

consultation for corporates, data-centers

Turnover rate is fast

All major cloud service providers are entering and offering solutions in this area

Reasons include better data privacy, backup policy could be controlled by corporations

Future: Public Cloud + Private Cloud => Hybrid Cloud

About ClusterTech

CLUSTERTECH was founded in 2000. CLUSTERTECH headquarters in the Hong Kong Science Park

CLUSTERTECH has 70 professional staffs with advanced degrees (16 Ph.D.s) in science and engineering and reputable clients in financial and government sectors. CLUSTERTECH is a one-stop provider offering high performance computing and business intelligence solutions, through in-house developed CPE, CGE, and CHESS platforms.

Client Lists

HSBC

Different artificial intelligence modelling techniques were developed to forecast capital flows and market directions in foreign exchange.

Hong Kong Observatory

Provided consulting and software implementation on its numerical weather forecast platforms

Ping An Insurance

Provided data cleansing and mining solution to Ping An Insurance on their insurance customer database

Business Case: Bank of China

Problem and Requirements: Fast access on market data

High computational job, lots of batch processing

Required system stability and high availability

Scale-up and scale-out with minimal admin cost

A customized solution with Clouds Architecture: Take advantage of our expertise in parallel computing

Users do not see the physical hardware layers, only “the cloud”

Empowers client with scalable software (SaaS) and infrastructure (IaaS)

Overall Cloud Applications

Application I: Low Latency Market Data Service

e.g. Reuters, Bloomberg, HKEX…

Application II: Computational Service

Application III: Parallel Monte Carlo Sim.

Application IV: Middle Office Data Backup

Business Case Summary

A customized solution with clouds architecture: Software

Low latency market data access An excel module to assign job for parallel monte carlo simulation on the

cloud for fair pricing

Infrastructure Users could assign computational intensive jobs to the cloud Users could take advantage of database stored in the cloud for safe

storage

END OF PRESENTATION

Email: [email protected]: (852) 2655 6158

Units 211-213, Lakeside 1, No 8 Science Park West Avenue,

Hong Kong Science Park, Shatin, NT, Hong Kong.