Leveraging Cloud Computing for
Disaster Recovery
November 18, 2010
Bill Russell
[email protected]
[email protected]
© Bick Group 2010
A GENDA
• What is Cloud
• How do we leverage it for disaster recovery
© Bick Group 2010
W HO WE ARE
© Bick Group 2010
OBLIGATORY CREDIBILITY SLIDE
#1
#1
#1
BEST SELLING
BOOK ON
CLOUD COMPUTING
MOST READ
BLOG ON
CLOUD COMPUTING
MOST LISTENED TO PODCAST
ON CLOUD COMPUTING WITH
OVER 10,000 LISTENERS
© Bick Group 2010
HYPE
Source: Gartner Research
“How Cloud Computing is Changing the World”
© Bick Group 2010
Primary Source
© Bick Group 2010
THE DALLES FACILITY
Undertaken just 5 years ago and located in the northern-most reaches of Oregon
along the banks of the Columbia River is what one writer has called
“the information-processing equivalent of a nuclear
power plant, a data dynamo of unprecedented power.”
© Bick Group 2010
CLOUD COMPUTING DEFINED
ESSENTIAL CHARACTERISTICS
Self Service
Provisioning
Rapid Elasticity
Metered Service /
Pay as you go
SERVICE MODELS
Infrastructure as
a service (IaaS)
Platform as a
service (PaaS)
Software as a
service (SaaS)
Resource Pooling
DEPLOYMENT MODELS
API
Broad Network
Access
Private
Cloud
Community
Cloud
Public
Cloud
Hybrid
Cloud
© Bick Group 2010
CLOUD ECOSYSTEM
THE "CLOUD PYRAMID"
• Building blocks: IaaS -> PaaS -> SaaS
Google
App Engine
Source: GoGrid
© Bick Group 2010
ECONOMICS OF THE CLOUD
With the cloud, you use and pay for only what you need.
Capital Expenditure
Not enough capacity,
increased costs and
dissatisfied clients
Capacity:
Compute,
Storage,
IT Labor,
Real Estate
Capital Expenditure
Capital Expenditure
Underutilized but still
have capacity expenses
Capital Expenditure
Traditional Provisioning
Cloud Provisioning
Actual Usage
Capital Expenditure
Time
© Bick Group 2010
HIGH VALUE TARGETS
—
HIGH %
—
—
—
Telecom
Banking
Healthcare
High-tech
IT COST
STRUCTURE
(IT infrastructure
spending as
a percentage
of overall
IT spending)
—
MED %
—
—
—
LOW %
—
—
—
Pharmaceuticals
Manufacturing
Insurance
Airline industry
Transportation
Benefits administration
Some government entities
EXISTING IT ENVIRONMENT (percentage of x86-based compute images)
© Bick Group 2010
DISASTER RECOVERY
Bick's experience is that because of the small
probability of many disasters client may have
insufficient disaster recovery plans. However,
because the consequences of actual disasters
are often business failure clients must not let
the small likelihood of a single or particular
event bias their judgment
© Bick Group 2010
OVERVIEW
High
System Failure
Power
Hardware
Network
Software
Human Error
Sabotage
Hack Event
PROBABILITY
Building Inaccessible
Chemical Spill
Isolated Weather Event
Fire
Pandemic
Toxic Event
Workplace Violence
Low
Building Loss
Airplane
Water / Fire / Explosion
Weather Event – Ice/Tornado
Natural Disaster
Utility Failure
High
I M PA C T
© Bick Group 2010
VALUE OF CLOUD AND DR
• Capital and OpEx
– NO DATA CENTER INVESTMENT
– YOU CAN TURN IT ON WHEN NEEDED
– THIS OPENS OPPORTUNITIES FOR
BUSINESSES THAT TYPICALLY COULD NOT
AFFORD A BACK-UP CENTER
– THE DOLLAR ESTIMATE IS THAT THE COST
IS ABOUT A FOURTH THAT OF
TRADITIONAL BACKUP SITES
– THE TRICK IS TO FIND THE CLOUD
COMPUTING PROVIDER THAT MEETS
YOUR DR REQUIREMENTS
© Bick Group 2010
VALUE OF CLOUD AND DR
• THE SECOND, IS THE SPEED
IN STANDING UP A SITE
– IN MANY CASES, NO NEED
TO PURCHASE AND TEST
HARDWARE AND
SOFTWARE
– YOU CAN BECOME A CLOUD
COMPUTING CUSTOMER
TODAY WITH A CREDIT
CARD
© Bick Group 2010
VALUE OF CLOUD AND DR
• FINALLY, MOST CLOUD-COMPUTING
SYSTEMS ARE UBIQUITOUS
– THUS, IF YOU HAVE ACCESS TO A
BROWSER AND THE INTERNET, THAN
YOU CAN ACCESS YOUR CORE
BUSINESS SYSTEMS AND CONTINUE
YOUR BUSINESS ANYWHERE IN THE
WORLD
© Bick Group 2010
METHODOLOGY
Identify Financial, Operational, and
Intangible impacts that a business interruption
would introduce
Identify interdependencies of the business processes,
applications, and technology systems
Establish recovery time (RTO)
and recovery point (RPO) objectives
for critical systems
Define priorities for re-instating
critical processes and systems based on the
requirements defined by the business
© Bick Group 2010
A NALYTIC R OADMAP
Assessment based on impacts to clients
Impacts
Business Impact
Analysis Survey
Priorities
Data Analysis
Performed
Requirements
Deliverables
Produced
Revenues
produced
from clients
by Business
Service
Client SLA,
contracts
and business
impacts
Client
applications
(client recovery
objectives for
applications)
Client
supporting
business
functions and
vendors
Financial
analysis on
client revenues
to define
priorities
Impact analysis
to define client
impact rating
Application RTO
& RPO
assessment
based on client
data
Application
recovery priority
analysis based
data and user
input
Financial
Analysis
Scorecard
Client
Impact Rating
Scorecard
Application
RTO/RPO
Scorecard
Application
Recovery
Priority
Matrix
© Bick Group 2010
OBJECTIVES
Application Recovery Priorities
Tier 1 (0-24 Hours)
Immediate
0-4 Hrs.
Tier 2 (1-3 Days)
4-24 Hrs.
1-3 Days
Tier 3 (> 3 Days)
3-7 Days
1-3 Weeks
>3 Weeks
Network
Infrastructure
Database
Infrastructure
Shared
Applications
Client Dedicated
Applications
General
Business/Office
Applications
© Bick Group 2010
DISASTER RECOVERY
Single Site
Multiple Sites
Active/Active
Active/Passive
Hybrid
Live or Standby
Level of resiliency
Synchronous
Asynchronous
Facilities
Function
Network
Data
SPECTRUM OF CHOICES
Ownership
Proximity
Parent or Company
Own or Lease
< 65 miles
> 65 miles
Systems
Live or Standby
Level of resiliency
© Bick Group 2010
FACILITY OWNERSHIP
PRIMARY ISSUES
SPECTRUM OF CHOICES
 Control
 Value
 Capital
A
A.
B.
B
Own
Lease
C. Recommendation
REASON #1
REASON #2
REASON #3
Control
Value
Capital
 Some control is lost in a
lease situation
 Owning is a bad investment
 Better Facility
 Less capital
 Owner will spend to keep
facility current due to
competitive pressures
© Bick Group 2010
“ TO BE”
© Bick Group 2010
CLOUD ECOSYSTEM
THE "CLOUD PYRAMID"
• Building blocks: IaaS -> PaaS -> SaaS
Google
App Engine
Source: GoGrid
© Bick Group 2010
A HIGH LEVEL PROCESS
1.
anddocument
document
the “As-is”
1. Understand
Understand
and
the “As1. Understand
and document
the “AsArchitecture
and
major
challenges
Is” Architecture
and major
challenges
Is” Architecture
and
major challenges
5. Define the “ToBe” Architecture
3. Evaluate
BusinessBusiness
Cases Cases
3. Evaluate
AAs-Is
S - I S Architecture
AAs-Is
R C HArchitecture
ITECTURE
To-Be
TTo-Be
O - B E Architecture
A R C HArchitecture
ITECTURE
of cloud platform
candidates
of cloud platform
candidates
3. Evaluate Business Cases of
cloud platform candidates
5. Define5.the
“To- the “ToDefine
Data
Business Business
ProcessesProcesses
Business
Cases
BUSINESS
CASES
Business
Cases
Business Business
ProcessesProcesses
Data
Applications
Applications
Infrastructure
Infrastructure
Residual Risks
Residual Risks
Data
4. Evaluate Security Risks of
cloud platform candidates
4. Evaluate
Security Security
Risks Risks
4. Evaluate
Infrastructure
Infrastructure
Security Architecture
Security Architecture
(Residual + New)
2. Select
businessbusiness
2candidate
. Selectbusiness
candidate
2. Select
candidate
Security
Risks Risks
Security
processes,
services,
processes,
services,
processes, services,
applications,
+ New)+ New)
(Residual
applications,
data, anddata, and(Residual
applications,
data, and
supporting
supporting
infrastructure
for
supporting
infrastructure
for
infrastructure
for cloud
platforms
cloud platforms
cloud platforms
Data
Applications
Applications
of cloud platform
candidates
of cloud platform
candidates
SECURITY RISKS
Be” Architecture
Be” Architecture
6. Develop
andand
migration
6. Develop
cost cost
andcost
migration
6. Develop
migration
plan,
cloud
vendor
analysis
plan, cloud
vendor
and
plan,
cloudanalysis
vendor analysis
and
and
evaluation
evaluation,
and finaland and
evaluation,
final final
recommendations
recommendations
recommendations
© Bick Group 2010
HIGH LEVEL
APPLICATION INSTANCE
Confocal
Xserv
Image Processing
PUBLIC / COMMUNITY IAAS CLOUD
APPLICATIONS
Workstation
APPLICATIONS
Workstation
Identity
Management
Workstation
Instance
Blade Server
(Access Control,
Accountability)
Workstation
Instance
Blade Server
Instance
SAN Storage
On-Premises
Security Controls
Tenant Separation
Storage
Xserv
Data Security
Hardened
Virtual Machine
Researcher
Centralized Resource
Provisioning, De-provisioning,
and Monitoring
© Bick Group 2010
FINAL WORDS OF ADVICE
Let the business
requirements lead you,
not the hype.
Leverage SOA
approaches
and best practices.
Make sure to consider
security and governance
as systemic concepts.
Never lose control of your data
Make sure to
consider performance
and scalability.
Understand the hard
and soft costs up front.
Understand all compliance
issues up front.
Don’t be afraid to start
over, if needed.
Learn all you can
before starting the project.
© Bick Group 2010
THANKS!
BLOGS
InfoWorld, Intelligent Enterprise, eBizq.net
W E E K LY P O D C A S T S
Cloud Computing Podcast
COLUMNS
SOA World Magazine, Cloud Computing Journal
Follow me on Twitter (BillatBML)
Follow Dave on Twitter (DavidLinthicum)
We Build Clouds.
Bill Russell
[email protected]
David Linthicum
[email protected]
© Bick Group 2010