• No results found

Cloud Computing. Carlos Passi, Assistant Controller, Business Transformation, IBM Corp. Steve Ford, CFO, TradeCard. March 24, 2011

N/A
N/A
Protected

Academic year: 2021

Share "Cloud Computing. Carlos Passi, Assistant Controller, Business Transformation, IBM Corp. Steve Ford, CFO, TradeCard. March 24, 2011"

Copied!
41
0
0

Loading.... (view fulltext now)

Full text

(1)

Cloud Computing

Carlos Passi

,

Assistant Controller, Business Transformation, IBM Corp.

(2)

Using Cloud computing to deliver innovation and efficiency

A new consumption and delivery model

Case Studies

TradeCard

IBM Development Cloud

Round Table

(3)

Using Cloud computing to deliver

innovation and efficiency

Carlos Passi

(4)

Cloud is:

– A new consumption and delivery model

Cloud addresses:

– Cost reduction

– Scale

– Utilization

– Self-service

– IT agility, flexibility and delivery of value

Cloud represents:

– The industrialization of delivery for IT supported

services

Cloud includes:

– Deployment models: public, private, hybrid

– Delivery models: Infrastructure as a Service (IaaS),

Platform as a Service (PaaS), Software as a Service

Cloud is a shift in the consumption and delivery of IT with the

goal of simplifying to manage complexity more effectively.

(5)

The operational benefits most organizations are looking for

in the cloud…

1.

Cost savings

IT and line-of-business decision-makers demand at least 20% cost savings

to move into the public cloud.

2.

Flexibility

After security, deployment flexibility is the biggest driver of cloud adoption

worldwide.

3.

Security

Nearly 25% of developers cite security as the primary inhibitor to cloud

adoption.

4.

Ease of Use

As use extends from IT to line-of-business leaders, and ultimately to every

employee, simplicity becomes increasingly critical.

5.

Scalability and Accessibility

As your organization grows, you need the computing capacity to match.

You also need access from anywhere in the world, at any time

(6)

There are three ways to acquire IT capabilities

Cloud computing is a new delivery and

consumption model or methodology

spanning all 3 ways.

Software, hardware

(7)

What is different about cloud computing?

With cloud computing

Without cloud computing

Virtualized resources

Automated service

management

Standardized services

Location

independent

Rapid scalability

Self-service

Software Hardware Storage Networking Software Hardware Storage Networking Software Hardware Storage Networking

(8)

Cloud computing delivers IT and business benefits

Automated

Faster cycle times

Lower operating expenses

Optimized utilization

Improved compliance

Standardized

Easier access

Flexible pricing

Reuse and share

Easier to integrate

Virtualized

Higher utilization

Economy of scale

benefits

Lower capital expense

Higher quality services

Doing more with less

Breakthrough agility

and reducing risk

(9)

Design a cloud computing deployment as part of the existing IT

optimization strategy and roadmap

Consolidate

Virtualize

Standardize

and automate

Reduce infrastructure

complexity

Reduce staffing

requirements

Manage fewer things

better

Lower operational costs

Remove physical

resource boundaries

Increase hardware

utilization

Reduce hardware

costs

Simplify deployments

Standardize services

Reduce deployment

cycles

Enable scalability

Flexible delivery

(10)

Changing the economics of IT Automating service delivery

Radically exploiting standardization Rapidly deploying new capabilities

IT-focused

An evolution of information technology

The impact of cloud computing is extending into driving

business transformation

Business-focused

Creating new business modelsEnabling speed and innovation Reengineering business process Supporting new levels of collaboration

An enabler of business transformation

Efficiencies

Transformation

(11)

Web infrastructure

applications

Collaborative infrastructure

Development and test

High Performance

Computing

...

Test for Standardization

Examine for Risk

Database

Transaction processing

ERP workloads

Highly regulated workloads

...

High volume, low cost

analytics

Collaborative Business

Networks

Industry scale “smart”

applications

...

Explore New Workloads

Workload characteristics determine standardization

(12)
(13)

Private

Public

Hybrid

IT capabilities are provided “as a

service,” over an intranet, within the

enterprise and behind the firewall

Internal and external service delivery

methods are integrated

IT activities / functions are

provided “as a service,” over

the Internet

Third-party

operated Third-party hosted and operated

Enterprise

data center data center Enterprise

Private cloud Hosted private

cloud Managed private cloud Enterprise Shared cloud services A Enterprise B Public cloud services A Users B • Free • Register • Credit Card • Click to contract

(14)

Database- and application-oriented

workloads emerge as most appropriate

Data mining, text mining, or other analytics Security

Data warehouses or data marts

Business continuity and disaster recovery Test environment infrastructure

Long-term data archiving/preservation Transactional databases

Industry-specific applications ERP applications

Infrastructure workloads

emerge as most appropriate

Audio/video/Web conferencing Service help desk

Infrastructure for training and demonstration WAN capacity, VOIP Infrastructure

Desktop

Test environment infrastructure Storage

Data center network capacity Server

Top public workloads

Top private workloads

(15)

Economic Drivers

Standardization

Automation

Scale

Utilization

Major Factors that Impact each

Driver

What workloads are you running?

– To what scale are those workloads?

– How varied are the utilization

patterns of the workloads?

Greenfield or Existing applications?

What are the software licensing Ts and

Cs for virtual vs. physical?

Are there internal policy / governance

issues that will limit the benefits of

cloud?

Are there legal / regulatory constraints

to leveraging shared infrastructure –

inside or outside the company?

What level of hardware and network

isolation are required for the workload?

If yes,

choose

deployment

model

ROI

ROI

Cloud

works?

(y/n)

(16)

Cloud Scale Economics: Simple Infrastructure Model

0 0.01 0.02 0.03 0.04 0.05 0.06 0 100,000 200,000 300,000 400,000 500,000 600,000

Enterprise Model

(Third Party SW)

Large Scale Cloud

Service Provider

Service providers have larger ‘fixed costs’ for developing the service platform - Point of inflection is at about 150K VMs 2x cost reduction from 160K to 600K VMx

Scale is an important factor to determining whether the economics make sense for

dedicated infrastructure - Point of inflection in cost for Enterprise is about 32,000 VMs

Large Scale Cloud Service Provider Enterprise with 3rd

Party Hypervisor

Number of VMs( 1 vCPU2GB memory & 160GB storage)

C

os

t p

er

V

M

H

ou

r

Large fixed costs to set up the cloud translate to the need to have sizable scale to reach a point of inflection on cost

(17)

$0.00 $0.05 $0.10 $0.15 $0.20 $0.25 $0.30 $0.35 $0.40 $0.45 $0.50 0 5,000 10,000 15,000 20,000 25,000 30,000

Enterprise Model

(Third Party SW)

Cloud Service Provider Price

Number of VMs( 1 vCPU2GB memory & 160GB storage)

C

os

t p

er

V

M

H

ou

r

Scale of the workload impacts decisions to deploy dedicated vs. procuring shared cloud models from a service provider.

There are large upfront costs to setting up cloud infrastructures, which require scale

This point of indifference varies by workload

(18)

Migration Costs Increase with Complexity

1. Standalone workloads -potential for migration to AIX/UNIX/Linux virtual machines via Bulk Movecapability

2. Easier workloads– also potentialfor bulk move capability. Complexity increases if multiple appsco-mingledacross server ecosystems

3. Medium complexity- due to size, security and risk. Managing connections to upstream & downstream applications is a must

4. Complex workloads– multiple applications & servers comprise the workload. All must move together to reduce Testrequirements. Lack of adherence to standards and versions

elongates migration

5. Very Complex- dependent on ISV’s investment in code running on that platform.

ERP applications typically require upgrades and vendor mandated tools & migration methodology. Testing is long pole

6. Completely custom– often requires remediation to resolve O/S specific calls,

1

2

3

4

5

6

1

2

3

4

5

6

Work load Migrat ion Diffic ult y

1

2

3

4

5

6

1

2

3

4

5

6

Work load Migrat ion Diffic ult y Work load Migrat ion Diffic ult y

When you have to move work, there are other factors: standardization & migration costs. There is a tradeoff between how much you spend to move, vs. the return you get in run

(19)

Security is among a top concern with cloud computing...

Data and information

Understand, deploy and

properly test controls for

access to and usage of

sensitive data

People and identity

Mitigate the risks

associated with user

access to corporate

resources

Application and process

Keep applications secure,

protected from malicious or

fraudulent use, and hardened

against failure

Network, server and end point

Optimize service availability by

mitigating risks to network components

Physical infrastructure

Provide actionable intelligence on the

desired state of physical infrastructure

security and make improvements

(20)

A systematic approach to cloud computing can ensure a

successful implementation.

Deliver

Plan

Build

Integrate public cloud services

Design and construct private clouds

Quality assurance (test)

Security and compliance

Lifecycle management

Understand strategic direction

Analyze workloads

Determine delivery model

Define architecture

Build the business case

Pilot

Deploy

Consume

Manage

(21)

In Summary

In the end,

scale and utilization dominate the economics

– you can only get

these through high degrees of standardization and automation

Greenfield

deployments have a simpler economic equation – because its

just

run costs that matter

Existing workload

migrations need to

weigh cost/complexity of migration

with run cost

benefits of standardization and automation – again, scale and

utilization matter

(22)
(23)

TradeCard is a Global Organization

Roughly 50% of workforce based overseas.

!!

"

(24)

(

)

*

+,

$ $

$ %

-

$ .

$

$

$

$

/

$

$

0 1 -2 ,

+

$

$

.

,

+3

(25)

4

$

%

$ 2 *

4

$

$ , .

%

!

5

%5

$

4

$

4

$

$

! $

4

$

$

)

%

(26)

4 6 !!

$

$

7

4

'

4

.

%

8

9 $ !!

4 (

$

:

,

+

4

%

(27)

4

#

4

$

.

% ;

, . +

+ . $

)

*(

4 6 !!

!

$

$

(28)

Hong Kong

Taiwan

Shenzhen

Sri Lanka

AD DNS DHCP File Serve Backup AD DNS DHCP File Serve Backup AD DNS DHCP File Serve Backup AD DNS DHCP File Serve Backup AD DNS DHCP File Serve Backup

(29)

-4

=

, .

4

% , .

$

4 *

,

.

4

!

4

4

,

'

%

4 > $

> +

Availability

Functionality

Productivity

Cost

Risk

Complexity

(30)

; >(

*

5 >7 5 * 5? >

/

'

;

=

%

6

5

-

(

0!

;

5

$

;

; @>5@

5 .

(31)

Development & Test Cloud in IBM

Carlos Passi

(32)

Agenda

Why Focus on Dev/Test?

The Story of Dev/Test

The Dev/Test Experience

IBM Results

New Uses of Dev/Test Cloud

Conclusion

(33)

w3Cloud

w3Cloud

Higher Gain

High amount of data transfer Long-running, stable demand Workloads amenable to standardization Significant application redesign to conform to cloud architecture Self-contained applications/ SOA

Legacy or highly complex services

High degree of workload customization/configuration

Easier

Delivery

Security & Business Resiliency Tradeoffs

Small scale or high volume

Storage Storage Dev/Test Dev/Test Notes Environ Notes Environ Analytics Analytics w3Cloud w3Cloud Desktop Desktop Blue Harmony Blue Harmony

Harder

Delivery

Higher Gain Easy Delivery Rapid Adoption Long/Short Term Increased ROI

Focus

(34)

Dev/test environments are provisioned manually via System Administrators; average 5 day turnaround Lower server utilization rates; 35% virtualization today

Before

Enable internal development teams to more quickly build and manage their development/test

environments based on several OS and middleware templates available via the Cloud.

The Project

Reduced provisioning time from 5 days to 1 hour Increased flexibility to de-provision images when no longer needed, freeing up system resources

Increased rate of virtualization and consolidation Reduced system admin labor spent building environments and deploying middleware products

After Development/Test Cloud

Time and cost savings resulting from self-service enables the hosting support team to handle significant growth, new clients, and innovations in

Our Development/Test Cloud enables lower cost hosting while

providing increased service for internal development

(35)

Dev/Test Streamlined

Administration Costs Dev Test Availability Infrastructure Usage Metrics Automated Billing Standardization Time to Market

Automation

End

Users Service Portal Service Request Catalog

Provisioning Engine

Workflows Scripts

Billing Features:

Usage Reports

- Base OS/Web Image - Memory - CPU - Disk - MW Support Service Increase Virtualized Infrastructure

(36)

IBM Results

.

Expected ~60% adoption. Realized ~100%

Custom configurations adopting standard offering Reduce non standard costs

Benefits virtualization strategy Reduced capital expenditures Increased consolidation

Decreased labor costs

Improved admin to server ratio

Other areas of use presented

Controller (Supplier) Perspective

Focus on high value transformation

Avoid red tape of server enablement Time to Value

Decreased cost of development (Time and Materials) Usage based costing

(37)

An innovative new way to source and

complete projects that uses competitions and free cycles to deliver value to projects…

With TopCoder/Liquid, professionals operate in an innovative peer-based model, and compete against others to deliver work on time that meets quality objectives.

.

GenO is a community-based collaboration model for business application development

With TopCoder/Liquid, the community model is extended, permitting project teams to access

subcontracted services to help with application design and development work through software competitions. The Dev/Test Cloud enables the GenO Liquid team to rapidly provision images on required platforms, with software needed to support coding competitions for TopCoder resources.

The Project

Allocation based billing for cloud services aligns to TopCoder model for short term computing needs (+/-week). Avoid paying for full month of hosting (e.g., 75% savings from traditional hosting)

Eliminate the cost of manual set up for each new images

Benefits using the Development/Test Cloud

New Use for Dev/Test - Using the Dev/Test Cloud as part of our

(38)

IBM has a tremendous optimization opportunity available to it by leveraging cloud

Making good progress in an evolving, maturing space

Exciting work with tremendous interest from many constituencies

We can’t go fast enough!

(39)
(40)
(41)

Carlos Passi

Assistant Controller, Business Transformation

IBM Corporation

Steve Ford

Chief Financial Officer

TradeCard

References

Related documents

Students in the middle school social studies concentration take the following courses in addition to the ETE General Studies and Professional Studies courses.. Course

Enterprise 
 Cloud Consumers Ethernet Service Provider On-Net Private Cloud Services Internet Public Cloud Services. Cloud Service Delivery via Carrier

Private cloud Hosted private cloud Managed private cloud Enterprise Shared cloud services Enterprise A Enterprise B Public cloud services A Users B.. IBM Smart

A Private Cloud Service Provider offers application hosting and software services available through a private cloud dedicated to a single enterprise.. A Public Cloud Service

However, when maintenance or repair services or parts are provided under those agreements, the service or repair companies will be acting as service providers under provisions

In spite of these weaknesses, our study of a small patient cohort showed a significant reduction in health care con- tacts caused by somatic disease incidents during OMT compared to

In this section, we use the result for Dirac-type systems to establish Theorem 1: The Cauchy data of a connection Laplacian plus potential on a surface with boundary determines

Pavement surface texture, characterized by microtexture, macrotexture, and megatexture, is a property used to describe the functional condition of pavements. It can be