Few terms are more confusing than the term "cloud." While we've moved beyond the age of "defining the cloud," there's still a significant amount of confusion in understanding the role of the public cloud in an enterprise IT infrastructure. This presentation defines the elements of a mature enterprise cloud computing strategy that includes all components of a hybrid cloud, how to build out an integrated public/private infrastructure, and strategies for when and where to deploy new systems, and when it makes sense to migrate existing systems.
2. The four stages of cloud adoption
1. Experimentation
2. Going “rogue”
3. The scramble for order
4. Enterprise adoption
3. Understanding the “rogues”
The controls in place in most IT shops have become a barrier to
business getting work done
Cloud computing has given more widespread and easily consumable
access to lower levels of technology to all business stakeholders
The “rogues” use the cloud to gain agility and economic freedom
4. The “rogues” can put the business at risk
Governance
Compliance
Risk management
5. An enterprise cloud strategy…
Preserves (and even enhances)
agility
Makes possible the graceful loss
of centralized control
Minimizes risk
6. Why cloud matters to the enterprise
On-demand, metered provisioning with the perception of
infinite capacity
- On demand
When the business needs it
- Metered
Pay for what is used, dispose of when done
- Perception of infinite capacity
No worries about resource constraints or location
8. How public cloud computing happened
A handful of companies have achieved
massive economies of scale.
Cloud computing gives an enterprise
on-demand access with the perception
of infinite capacity to the economies of
scale realized by massive scale
organizations like Amazon.
9. What makes the cloud so easy?
Credit card + need = consumption
Limited (or even no) technical
knowledge required
Immediate gratification
10. Why is enterprise adoption so hard?
No single approach serves all cloud
computing needs
Using cloud effectively requires an
understanding of “the whole cloud”
Private clouds are technically complex and
cross many IT/business disciplines
IT must be involved to manage risk across
the enterprise
11. Elements of a cloud strategy
Cloud vision:
what cloud means to the organization
as a whole
Service models: IaaS, PaaS, and
SaaS
Delivery models: public and private
The cloud computing stack
13. Before defining your cloud strategy
You should have some practical experience (dev/test
counts) in an IaaS like AWS
Ideally, some PaaS experience
You should have a solid appreciation of the value of
cloud computing
You should understand the risks that arise from different
cloud models
No unicorns, no FUD
14. The cloud vision
A statement that describes the value of cloud in
the abstract to the enterprise as a whole
- Business and IT should agree on this vision
- Don’t spend too much time defining a vision
—Better to execute
Align processes and controls with the vision
Migrate to the cloud to realize this vision, not
simply for the sake of “going cloud”
17. The benefits of different service models
SaaS is the ultimate commoditization of well-understood software problems,
but poorly suited to bespoke systems or other custom needs; often poorly
suited to regulated environments.
PaaS is the ideal cloud environment for rapid development of bespoke
software systems that require nothing specific from the underlying
infrastructure; the focus is on apps. Poorly suited for regulatory and other
compliance needs or any but the simplest infrastructure needs.
IaaS provides the maximum control over infrastructure without the need to
worry about infrastructure; well suited to complex, custom software systems.
Network and storage I/O can be a challenge for IaaS.
Virtualized data centers and physical infrastructure still exist and have a role
for systems with very specific, non-commodity hardware requirements.
18. Delivery models
Public cloud is a multi-tenant cloud infrastructure
delivered by a provider that has achieved scale (IaaS,
PaaS, and SaaS)
- For the greatest level agility
Private cloud is a single-tenant virtualized
infrastructure with consumption characteristics similar
to a public cloud (IaaS and PaaS)
- For greater control for regulatory, compliance,
or bespoke requirements
21. Selecting the right deployment model
Software needs will migrate from private/physical to public/SaaS over time
due to commoditization
When selecting a deployment model for today, you need to answer two
questions:
- How commoditized is my problem? (e.g. what level of service model
is appropriate)
- What are the governance and compliance needs? (e.g. what
deployment model)
Pick the service/delivery model combination that best matches your answers
- Bias towards public
- Bias towards SaaS
- The reality is that public/IaaS is the sweet spot for most bespoke
needs
Where you are in the overall adoption of cloud within your enterprise will
impact which projects should go first
22. Cloud Strategy: The Public Cloud
Don’t be fooled by the term “enterprise cloud”
- You don’t achieve “enterprise” by destroying
the value of “cloud”
- A public cloud is “enterprise” if it can help
you realize your cloud vision across a class
of business needs
Public clouds can be secured for many needs
- even sensitive data
- only specific controls should eliminate the
consideration of public for a specific problem
23. Time to provision matters MUCH more
than contracted service levels or
availability
24. Cloud Strategy: The Private Cloud
The primary goal of your private cloud is to
create a cloud-like experience capable of
meeting specific control and application
architecture objectives
Virtualization is NOT sufficient for a cloud-like
experience
Virtualization + cloud platform = private cloud
25. Cloud Strategy: Governance
Proper cloud governance enables you to minimize risk and
ensure compliance across a hybrid cloud infrastructure
Unified view of all resources in all clouds
Audit trail across all clouds
User access controls aligned with job function across all clouds
Integration with enterprise systems management tools
27. Cloud Strategy: Legacy and Greenfield
The biggest architectural issue is how to achieve fault
tolerance
- Some clouds are better for the more resilient “design
for failure” model
- Others are better for the traditional VM migration model
New applications should be created under the “design for
failure” model using distributed databases managed through
automated cloud management tools orchestrating the behavior
of next generation configuration management tools
But tools like Amazon RDS, Amazon SQS, etc. make it easier
to leverage more traditional technologies
And you may still want to migrate existing applications into the
cloud to leverage its flexibility
28. After the cloud strategy…
The best first step in execution is a structured proof
of concept
Go greenfield for the PoC, don’t migrate
Create measurable, achievable objectives
Don’t tackle too many new things at once
Make it comprehensive, include virtualization, cloud
platform, cloud management, configuration
management, public cloud
Success factors must include buy-in from the
“rogues”!