Sample of workshop given at CloudAsia 2012. Workshop is 700 slides, so this is just a small sample to give a feel for the content, depth and independent approach.
2. Outline: Part 1: Introduction to Cloud Computing
• Confusion and Cloud-Washing
• Cloud Consolidation
• History
• Vision
• Definitions – focus on NIST
• Cloud computing reference architecture
• Actors, Brokers, Consumers, Auditors,
• Cloud Types: Public, Private, Community and Hybrid
• Orchestration and Management
• Business support, security and privacy
• Cloud Benefits and Issues
• Cloud Misconceptions
• The Open Group Survey 2011
2
3. Outline: Part 2: Getting into the Details
• Mapping suppliers and technologies in Cloud Computing
• Understanding the economics behind the benefits
• Quantifying the benefits
• Cloud market taxonomy and market size
• CSPs and Cloud Computing
o AT&T, BT, DT, NTT, Orange, SingTel, Verizon
• Mapping the workloads
• SOA and the Cloud
• Cloud Computing in Asia
3
4. Outline: Part 3: Understanding the Components
• Summary: Web 2.0, SaaS, Utility Computing, Virtualization, SLAs, Autonomic computing, Grid
technology, Web Services, Service Oriented Architectures, Free and Open Source Software
• Deep Dive: Virtualization
o History
o Issues and Trends
o Supplier review: Citrix, IBM, Linux, Microsoft, Novell, Symantec, Oracle, VMWare
• Deep Dive: Data Centers
o History and the drive for efficiency and availability
o Changes and pressures on DC – drive for DC management
o Capex and opex DC costs
o DC economics drives cloud computing
• Deep Dive: Force.com, Google, Microsoft and Amazon
o Force.com
o Google App Engine
o Microsoft Azure
o Amazon Web Services
• Netflix deep dive
• AWS walk-through
4
5. Outline: Part 4: Implementation
• Survey - what workloads others are moving into the cloud
• Summary
o Key points in cloud migration
o Industry : Workload : Cloudability Space
• Project Plan – example from IBM
• Decision Tree for implementing Cloud Computing
o The Open Group decision tree
• Security
o Reviewing SAS70, PCI DSS, ISO27001, NIST, HIPAA, FISMA, CoBIT, Data
Protection Directive, practical aspects
• Architectural Review
• Concluding Remarks
5
7. Outline: Part 1: Introduction
• Confusion and Cloud-Washing
• Cloud Consolidation
• History
• Vision
• Definitions – focus on NIST
• Cloud computing reference architecture
• Actors, Brokers, Consumers, Auditors,
• Cloud Types: Public, Private, Community and Hybrid
• Orchestration and Management
• Business support, security and privacy
• Cloud Benefits and Issues
• Cloud Misconceptions
• The Open Group Survey 2011
7
11. We Live in Hyped Times!
• “Amazon and PSN outages won't halt cloud revolution.” source The Register
• “SURVEY: Future-proofing the cloud.” source Network World
• “Virtualization, cloud computing to dominate Interop.” source Network World
• “Is Your Data Center Ready for Cloud Computing?” source Web Buyers Guide
• “Demystifying the Cloud – A Conversation with Dell’s CIO and CTO!” source Baseline Briefing
• “Cloud-enabled Wi-Fi: Less Dollars, More Sense” source Network World
• “Apple’s new services are expected to include a "digital locker" solution enabling consumers to
store their iTunes music, movie and television libraries on Apple servers for access on multiple
iOS-based devices.” source Fierce Mobile Content.
• “Brocade Unveils CloudPlex cloud architecture, an open framework for building virtualized data
centers, and offered a look at new technologies coming up in the near future to help make such
data centers possible. “ source CRN
• “CenturyLink goes from local to global player with Savvis acquisition.” source Fierce
Free Software Foundation founder Richard Stallman called cloud computing,
“worse than stupidity.”
Bottom-line: If you’re systems are down or you loose customer data its not the Cloud
Provider that suffers / goes out of business – they just issue a credit for the disruption. 11
12. First Phase of Cloud Consolidation
• Verizon acquired Terremark, a Infrastructure / Platform as a Service (I/PaaS)
provider, for $1.4 billion, to provide IT infrastructure services targeting the
enterprise market.
• Dell spent more than $2 billion in six months acquiring cloud technologies,
including PaaS provider Boomi, and is investing another $1 billion in a group of
global data centers.
• IBM acquired Cast Iron, Boomi’s competitor.
• Time Warner Cable acquired NaviSite.
• CenturyLink acquired Savvis
• Microsoft and Toyota forged a strategic partnership to build a global platform
for Toyota Telematics Services using Windows Azure.
• CA Technologies and Unisys entered into a joint venture that combines CA’s
virtualization and service management products with Unisys’ virtualization and
cloud advisory, planning, design and implementation services.
Likely see further consolidation as Telcos realizes their weaknesses in selling Cloud into
12
enterprise – particularly small medium enterprise
13. Evolution
• Cloud computing has evolved through a number of
phases which include grid and utility computing,
application service provision (ASP), and Software as a
Service (SaaS).
• But the overarching concept of delivering computing
resources through a global network is rooted in the
sixties.
Those
Sixties!!!
13
15. • One of the first milestones for cloud computing was the
arrival of Salesforce.com in 1999, which pioneered the
concept of delivering enterprise applications via a simple
website.
15
16. • The next development was Amazon Web Services in
2002, which provided a suite of cloud-based services
including storage, computation and even human
intelligence through the Amazon Mechanical Turk.
16
17. • Then in 2006, Amazon launched its Elastic Compute
cloud (EC2) as a commercial web service that allows
small companies and individuals to rent computers on
which to run their own computer applications.
17
18. • Another big milestone came in 2009, as Web 2.0 hit its
stride, and Google and others started to offer browser-
based enterprise applications, though services such as
Google Apps.
Purely
representational
purposes only!
18
19. • 2009 also saw the advent of Microsoft into the cloud
computing game with its product Windows Azure
• Azure as an operating environment "designed to manage
extremely large pools of computational resources." The simple
explanation is that Microsoft wants customers to run their
Windows-based applications over the Internet using
Microsoft's data centers, with Azure being the system that
organizes resources and handles spikes in demand.
19
20. And Now…….
• Many IT professionals recognize the benefits cloud
computing offers in terms of increased storage, flexibility
and cost reduction
• Considerations such as security, data privacy, network
performance and economics are likely to lead to a mix of
cloud computing centers both within the company
firewall and outside of it
20
21. The Dream of Cloud Computing
Integrated Circuit Utility Computing
Foundries
• Semiconductor Fabs Expensive • New Datacenters Very Expensive
– Typically > $1 Billion – Only a Few Companies Can
– Too Much for Most Designers Afford Huge Datacenters
• Fabs Take Outside Work • Utility Computing Datacenter
– Fabs Amortize Cost Owners Amortize Costs
– Other Designers Make Chips – Utility Computing Users Get
Advantages of Elasticity
• Allowed Explosion of Designs
– Datacenter Resources Shared
– More Players Afford Rented Fab Across Many Users
But a private cloud doesn’t deliver scale? 21
22. The NIST Definition of Cloud Computing
o Cloud computing is a model for enabling convenient, on-demand network access to a
shared pool of configurable computing resources (e.g., networks, servers, storage,
applications, and services) that can be rapidly provisioned and released with minimal
management effort or service provider interaction. This cloud model promotes availability
and is composed of five essential characteristics, three service models, and four deployment
models.
Characteristics
1. On-demand self-service Service models
2. Broad network access 1. Cloud Software as a Service (SaaS)
3. Resource pooling 2. Cloud Platform as a Service (PaaS)
4. Rapid elasticity 3. Cloud Infrastructure as a Service (IaaS)
5. Measured service
Deployment models
1. Private cloud
2. Community cloud
3. Public cloud
4. Hybrid cloud
22
23. Why Now?
From T-Systems, who has delivered SAP dynamic services since 2004
23
24. NIST 3 Cloud Service Models
• Cloud Software as a Service (SaaS)
o Use provider’s applications over a network
• Cloud Platform as a Service (PaaS)
o Deploy customer-created applications to a cloud
• Cloud Infrastructure as a Service (IaaS)
o Rent processing, storage, network capacity, and other fundamental computing
resources
• To be considered “cloud” they must be deployed on top of cloud
infrastructure that has the key characteristics
24
25. Service Model Architectures
Cloud Infrastructure Cloud Infrastructure Cloud Infrastructure
IaaS Software as a Service
PaaS PaaS (SaaS)
SaaS SaaS SaaS Architectures
Cloud Infrastructure Cloud Infrastructure
IaaS Platform as a Service (PaaS)
PaaS PaaS Architectures
Cloud Infrastructure
IaaS Infrastructure as a Service (IaaS)
Architectures
25
26. NIST Reference Model: Background
• The NIST cloud computing definition is widely accepted as a valuable contribution toward providing a
clear understanding of cloud computing technologies and cloud services.
• It provides a simple and unambiguous taxonomy of three service models available to cloud consumers:
cloud software as a service (SaaS), cloud platform as a service (PaaS), and cloud infrastructure as a
service (IaaS).
• It also summarizes four deployment models describing how the computing infrastructure that delivers
these services can be shared: private cloud, community cloud, public cloud, and hybrid cloud.
• Finally, the NIST definition also provides a unifying view of five essential characteristics that all cloud
services exhibit: on-demand self-service, broad network access, resource pooling, rapid elasticity, and
measured service.
• These services and their delivery are at the core of cloud computing. In the cloud computing model, the
primary focus is a more economic method of providing higher quality and faster services at a lower cost
to the users.
• In the traditional IT service delivery model, there is a large emphasis on procuring, maintaining and
operating the necessary hardware and related infrastructure. The cloud computing model enables CIOs,
IT project managers and procurement officials to direct their attention to innovative service creation for
the customers.
26
27. NIST Reference Model: Background
• The NIST cloud computing reference architecture focuses on the requirements of
“what” cloud services provide, not a “how to” design solution and implementation.
• The reference architecture is intended to facilitate the understanding of the
operational intricacies in cloud computing.
• It does not represent the system architecture of a specific cloud computing system;
instead it is a tool for describing, discussing, and developing a system-specific
architecture using a common framework of reference.
• The design of the NIST cloud computing reference architecture serves the following
objectives:
o to illustrate and understand the various cloud services in the context of an overall cloud
computing conceptual model;
o to provide a technical reference to USG agencies and other consumers to understand, discuss,
categorize and compare cloud services; and
o to facilitate the analysis of candidate standards for security, interoperability, and portability
and reference implementations.
27
28. NIST Cloud Computing Reference Architecture
• The NIST cloud computing reference architecture defines five major
actors:
o cloud consumer,
o cloud provider,
o cloud carrier,
o cloud auditor and
o cloud broker.
• Each actor is an entity (a person or an organization) that participates in
a transaction or process and/or performs tasks in cloud computing.
• A cloud consumer may request cloud services from a cloud provider
directly or via a cloud broker.
• A cloud auditor conducts independent audits and may contact the
others to collect necessary information.
28
32. Benefits
• Shorter provisioning times: The provisioning of servers, applications, and application
environments is far quicker and cheaper to do leading to quicker time-to-market for new
products and services, shorter project timescales, and faster benefit realization.
• Reduced capital outlay: The ability to buy computing resources, whether applications or
infrastructure on a pay-as-you-go basis reduces the need for capital investment in hardware
and software. This in turn may make it easier to finance projects, which can rely upon
revenue generation to finance project outlay far sooner than would otherwise be the case.
The burden of upfront investment and thereafter capital depreciation and the risk of
stranded investments should a project fail is reduced.
• Allows more use of “try before you buy”: The ability to try a new product or service is
enhanced through the use of Cloud Computing services where the investment in trials and
proof-of-concept activities is much reduced. Trialing also reduces the risk of later
implementations.
• Reduces the cost of “one-off” activities: One-off activities which would otherwise be
extremely costly to finance with purchased or traditionally leased computing resources can
be more cheaply provisioned using Cloud Computing; e.g., migration or data
cleansing/conversion activities. 32
34. SaaS is not dependent on PaaS which is not dependent
on IaaS – They’re independent
• This illustration implies a relationship between IaaS, PaaS, and SaaS and gives rise to the idea that
the three service models are necessarily layered one upon the other. Although both software and
platform services will rely upon some elements of infrastructure (the fundamental “plumbing” of IT;
e.g., servers, network, storage), to infer that all SaaS is founded upon a PaaS and that in turn upon
IaaS is an extrapolation which will not stand closer analysis.
• Were this true, then for the service model and characteristics of Cloud Computing to apply then
each layer would have to be separately deliverable as a service with all the attendant components
allowing metering, account management, billing, self-service, etc.
• In reality, in a given purchase or consumption of Cloud Computing
services, the interaction is with one of these layers. One is either
buying or consuming software, platform, or infrastructure. That the
means by which the provision of this service is achieved is invisible
and of no concern is one of the founding concepts of Cloud
Computing. Although it is tempting to assume that all sellers of SaaS
services have reached extremely high levels of maturity in their
provision of infrastructure, that they employ sophisticated and
highly effective virtualization, for example, may not actually be the
case. At the level of service interaction of a consumer of SaaS it will
not be apparent and nor should it be. 34
41. Outline: Part 2: Getting into the Details
• Mapping suppliers and technologies in Cloud Computing
• Understanding the economics behind the benefits
• Quantifying the benefits
• Cloud market taxonomy and market size
• CSPs and Cloud Computing
o AT&T, BT, DT, NTT, Orange, SingTel, Verizon
• Mapping the workloads
• SOA and the Cloud
• Cloud Computing in Asia
41
42. One More Look at the Cost Model
How Much You
Make Total in a “Pay How Much You Make The Compute Cost
as You Go” Cloud Per User Hour in a of the Work in a
“Pay as You Go” Cloud Datacenter
UserHourscloud × (revenue – Costcloud) ≥
Costdatacenter
UserHoursdatacenter × (revenue – )
Utilization
But You Pay for the Whole
Utilization Datacenter Even When It Is
Assumptions Make How Much You Make Underutilized!
a Big Difference in Total in a Datacenter
the Costs of Cloud Implementation of
Your App Have to Increase the Charge for
versus Datacenter! the Work You Do to Make Up for
Underutilization
42
44. Cloud’s goal: flip this equation
30% 70%
On-Premise Your Managing All of the
Infrastructure Business “Undifferentiated Heavy Lifting”
Configuring
Cloud-Based More Time to Focus on
Your Cloud
Infrastructure Your Business
Assets
70% 30%
44
45. Companies have different motivations for leveraging cloud
Analytics & Time to Value Employee Risk &
Security Productivity Compliance
Operations support 9
major commands, Creates an Enable collaboration 34,000-employee
nearly 100 bases, & ecosystem for PayPal across 300K global bank deploying a
700,000 active military 3rd Party developers employees as well as its private cloud from
personnel around the network of customers, IBM to centralize
world. Design secure Reduces developer partners and suppliers. management of
cloud infrastructure for effort to deploy a work Saving 30 minutes per desktops via an
defense & intelligence environment with day or 120hr per year enterprise class data
networks; insights seamless PayPal Test per person. center rather than at
about cyber attacks, Sandbox access
the user stations,
network, system or IBM LotusLive has 18 Gets greater remote
application failures, million users in 99 flexibility without
while automatically countries sacrificing control to
preventing disruptions. improve efficiency.
45
46. IBM Cloud Business Model
ROI Analysis Impact:
Reduction of Total Cost of Ownership of
Data Center Infrastructure
New
100% Development Liberated Reduced Capital Expenditure
funding for - Improved utilization reduces requirement for
Software new new capital purchases
Costs development, Strategic
transformatio Change Reduced Operations Expenditure
n investment Capacity - Lower facilities, maintenance, energy, IT
Power or direct service delivery and labor costs
Costs
saving
Additional Benefits
Deployment (1- - Reduced risk, less idle time, more efficient
Curren time) use of energy, acceleration of innovation
t IT Labor Costs
projects, enhanced customer service
Spend (Operations Software
and Costs
Maintenance)
Business Case Results
Power Costs Hardware,
labor &
Annual savings: $3.3M (84%)
(88.8%)
power from $3.9M to $0.6M
Hardware Labor Costs savings
Costs ( - 80.7%) reduced
Payback Period: 73 days
(annualized) annual cost
Hardware Costs of operation Net Present Value (NPV): $7.5M
( - 88.7%) by 83.8% Internal Rate of Return (IRR): 496%
Note: 3-Year Depreciation Period with 10% Discount Return On Investment (ROI): 1039%
Rate
47. Full Cloud Taxonomy
Level Of
Sharing
Public IaaS PaaS SaaS BPaaS PURE
Cloud CLOUD
@ Global MARKET
Provider
Virtual
Private Dynamic Integration- Dynamic Dynamic
Cloud Infrastructure as-a-Service Apps BPO
@
Dedicated
Services Services Services
Provider
EXTENDED
CLOUD
Infrastructure Middleware Apps BP MARKET
Private
Cloud
Virtualization Virtualization Virtualization Virtualization
@ In-house Tools Tools Tools Tools
Data Center
Infrastructure Middleware Applications Business Business
Processes Value
47
48. Cloud market size 2012
Level Of
Sharing
Public
PURE
Cloud
@ Global
~$15b Market CLOUD
Provider
MARKET
Virtual
Private
Cloud
@ ~$28b Market
Dedicated
Provider
EXTENDED
CLOUD
Private MARKET
Cloud
@ In-house ~$11b Market
Data Center
Infrastructure Middleware Applications Business Business
Processes Value
48
49. Cloud market size 2012
Cloud Service Master Data Change Availability
Level Of Management Management & Config & Performance
Sharing
Billing
Public IaaS PaaS SaaS BPaaS Security
Cloud
@ Global
Provider $1.5b $500m $12.5b $500m
Virtual Service
Private Dynamic Integration- Dynamic Dynamic Desk
Cloud Infrastructure as-a-Service Apps BPO
@
Services Services Services
Dedicated Provisioning
Provider
$8.3b $300m $6.2b $13.2b
Infrastructure Middleware Apps BP
Private
Cloud
Virtualization Virtualization Virtualization Virtualization
@ In-house Tools Tools Tools Tools Backup
Data Center $3.1b $300m $4.5b $3.1b & Recovery
Infrastructure Middleware Applications Business Business
Processes Value
Purpose is to demonstrate the roles cloud computing plays and current market size
49
50. Cloud Services as a % of IT
Worldwide IT Spending by Consumption Model
600
IT Cloud Services
On-Premise IT
500 10%
Worldwide IT Spending ($ billion)
44
400 5%
17
300 CAGR
26% 416
4%
200 359
100
0
2009 2013 50
51. Cloud Services Growth Impact
Sources of Incremental IT Spending Growth in 2013
Cloud vs. Traditional Products
485
460.4
480
Net new IT growth
Worldwide IT Spending ($ billion)
475 27%
= $27.3 billion
470
IT Cloud Services growth
465
460
73% Traditional IT product growth
455
IT Cloud
433.1
450
445
440
435
430
2012 2013
51
52. Main topics to be addressed prior adoption of cloud
computing paradigm
• Security
• Availability
• Performance
• Interoperability
• Flexibility
• Personalization
• Unit costs
• One time transition costs
• Total cost of ownership
• SLA stipulations
• Liabilities of the provider
• Lock-in risks and implications
It’s the same as any Service Provider Decision: Don’t get locked-in
52
54. Telcos in the Cloud
• Telcos committed US$11 billion to cloud pursuits in 2011
o Eight out of 10 transactions involve datacenter assets
• Service differentiation remains poor
o 122 new services, 70% mass-market offers, heavy SaaS usage
• Telco strengths are underplayed
o Security and cloud mobility aren’t creating an unfair advantage
• ROI is a long game….
o Cloud services contribute a single-digit percentage of telco revenues today
• Few have solve the problem of enterprise credibility
54
61. T-Systems Cloud Positioning
T-Systems has created
significant thought
leadership collateral in the
Cloud Computing space.
Its positioning of Cloud
computing has received
broad endorsement, its
Dynamic Enterprise Cloud
positioning has won it
significant business in
Germany.
If offers end to end SLAs,
from the desk top to the data
center. While other
operators have struggled to
make that end to end offer T-
Systems was one of the first
(in Germany anyway.)
“In Germany we are the only
provider to offer cloud
services with an end to end
SLA.” source VP Networks
61
63. Keane provides extensive IT credentials in SAP and Oracle implementations across many
industries as well as across the enterprise application stack. 63
64. Keane becomes the face of NTT Data America, the solutions group within NTT Data aims to achieve common
64
solutions across regions, though the mobile link remains weak.
65. Intelligroup has extensive SAP and Oracle implementation experience in Pharmaceutical, manufacturing and
consumer goods verticals. 65
66. Value Team is an Italian IT Consultancy, that is also strong in LATAM, again buying IT market share. Deal was
66
announced in April 2011. With this acquisition NTT Data now has solid global coverage.
67. How the NTT Groups Fits Together
Dimension Data focuses on deployment (of communication platforms – Cisco and Microsoft) and
maintenance of IT systems. NTT Communications focused on transport services. NTT Data
focused on IT Services. However, in practice there are many overlaps in Europe, Latin America
67
and North America.
74. Verizon Buys Terremark
• In January Verizon announced plans to acquire Terremark Worldwide for U.S.$1.4 billion or U.S.$19
per share in cash—an acquisition price that is four times Terremark’s projected 2011 revenue of
U.S.$351 million.
• The acquisition highlights the unique market dynamics of cloud computing. Not since the dot-com
boom has a market seen such explosive growth in startups together with rapid consolidation and
acquisition. It’s a land grab, and Verizon just bought a big chunk.
• In September 2010, Verizon entered into a partnership with Terremark that focused on the SMB
segment. Verizon’s Computing as a Service (CaaS) SMB runs on Terremark’s infrastructure and data
centers, but Terremark also has a strong presence in the large enterprise and federal government
segments.
• The acquisition instantly gives Verizon a stronger position in the enterprise cloud computing market.
• The acquisition is also good news for enterprises, because those that want to adopt cloud computing
services now have more and better options.
• Also in January Hewlett-Packard announced its HP Enterprise Cloud Services-Compute, a service that
expands its offerings and enables it to position hybrid cloud to enterprises.
Verizon’s acquisition of Terremark demonstrated VZB was committed to dominating the
Cloud Computing business . In part VZB was struggling with CaaS in gaining broader
market adoption – Terremark solves this issue. 74
75. Verizon Benefits
• Data center scale and build-out skills.
o Terremark has 567,000 square feet of data center space available as of Q2 2011, with
significant room to grow. More importantly, Terremark knows how to build data centers
that are uniquely able to deliver cloud computing services to enterprises.
• Growth in managed services.
o Fifty percent of Terremark’s business is higher margin managed services, including
enterprise cloud computing services.
• Strong security skill set.
o According to Yankee Group’s Anywhere Enterprise: 2010 U.S. Cloud Computing
FastView Survey, security is still one of the leading barriers to enterprise adoption of
cloud computing (see next slide).
o Coupled with Verizon’s acquisition of CyberTrust, Terremark’s Federal Information
Security Management Act (FISMA)-compliant data centers and best-of-breed cloud
security expertise give Verizon meaningful competitive differentiation.
VZB now has the best security credentials of any cloud based service provider. If it can
persuade the broader market of these credentials it has the ability to dominate the global
market. 75
80. Defining the Map
• Start by grouping enterprise applications into classes of
applications.
• Then depending on the lifecycle ( e.g. Test & Development, Staging
or Production) , usage environment and security requirements of a
class of applications, an enterprise architect can define a set of
principles and guidelines to help decide when to use cloud
computing service and what type of service to use.
• Next slide shows an example from an enterprise architect of a well
known global brand.
80
81. Example of One Enterprises’ Mapping
Class of Enterprise Test & Staging Production
Applications Development
Business Virtual Private Virtual Private Cloud Private Cloud
Communications Cloud
CRM (e.g. SAP, Public Cloud Virtual Public Cloud Public Cloud
Salesforce.com)
Applications Public Cloud Virtual Private Cloud Private Cloud
supporting critical
business processes
Productivity Public Cloud Virtual Private Cloud Public Cloud
Improvement
Financial Virtual Private Virtual Private Private Cloud
Cloud Cloud
81
82. Reality Check: Its not just security
• One of the main barriers to enterprise adoption of cloud computing
service is the effort required to migrate corporate applications from
their internal hosting service to public cloud or virtual private cloud
and vice versa.
• Technology such as Verizon CloudSwitch service is now available to
allow an enterprise user to seamlessly switch applications between
their internal (e.g private cloud ) hosting service and Virtual Private
Cloud or Public Cloud.
• This type of technology should help drive down the barrier to future
user adoption of third party provided cloud computing service.
82
83. Cloud Migration Reality Check Part 1
• Standalone web applications built specifically for a particular cloud can be rolled out
quickly and relatively easily using templates offered by the cloud provider or software
from third parties.
• But it’s far more complex to run an enterprise application in a preferred public cloud
while staying integrated with the internal environment and its associated services,
processes, tools, and relationships.
• Moving an enterprise application to a cloud takes extensive manual configuration,
complex engineering, and trial and error — with success not always assured.
A whole landscape of specifications for OS versions, storage, networks, and
management tools has to be mapped and modified for an external environment that
is usually unfamiliar to internal IT staff.
• In addition, the applications almost always need to reach back to services and
processes in the data center, setting up a number of integration issues that are not
easily resolved. Thus, migration projects often take weeks or longer, preventing many
companies from even considering cloud deployment.
83
84. Cloud Migration Reality Check Part 2
• The separate, largely isolated environment imposes management
challenges that don’t occur internally when the application is under
enterprise control.
• These same challenges also apply to new enterprise applications
developed in the cloud since they also require integration with data
center tools, processes, and services.
• Everything from authentication and authorization to internal
databases and basic services has to be managed separately for an
application to run in the cloud.
84
85. Source of Cloud Migration Delay and Blocking
• 1. Rebuilding the application stack within the cloud
• 2. Setting up the network
• 3. Adding end-to-end security
• 4. Managing the application in a separate environment
85
86. 1. Rebuilding the Application Stack
• The cloud has a model similar to a virtualized data center environment where users or administrators can
provision virtual resources such as CPU, memory, and storage from a pool of physical resources.
• However, the processes used for building, launching, and managing servers in a public cloud are very different
from those used internally.
• Most cloud providers today require you to start from one of their base templates. These templates are customized
for a particular environment, including tools, drivers, agents, or specific configurations for leveraging the available
networking and storage capabilities.
• Even when a provider offers a method for uploading existing application images, the drivers, tools, and
modifications associated with an application must be included for compatibility with the chosen cloud
environment.
• This creates a different starting point and will affect how application stacks are built and maintained.
• Using the cloud requires that these components be rebuilt to match the cloud provider’s environment. Many
applications take advantage of services that exist within a data center, such as DNS or LDAP, that are not available
in the cloud.
• This requires re-architecting the applications that depend on these services, whether duplicating the services
within the cloud, building methods to extend existing services to the cloud, or some combination of the two.
• These differences between the data center and the cloud trigger a chain of integration issues including potential
changes in base operating systems, storage, networking, virtualization, and shared services.
86
87. 1. Rebuilding the Application Stack: OS
• The cloud provider will specify operating system versions as well as versions for
related components such as storage and network devices, drivers, and
virtualization tools.
• However, complying with their requirements can be problematic. For example,
in Linux environments, cloud providers require a particular kernel version
which must be matched by any application-specific kernel modules.
• This is particularly difficult when using third-party software since the required
code and/or tools may not be available to allow recompilation.
• The hypervisor version also has to match, as do the drivers and tools which
interact with it. Conflicts may not be easy to resolve — for example, if a cloud
provider is using VMware ESX 3.5, and the enterprise has already moved to ESX
4.0.
87
88. 1. Rebuilding the Application Stack: Storage
• Storage and data management challenges in the cloud result from the diverse and often
unfamiliar processes offered by cloud providers.
• Cloud providers define the relationship between servers and storage, and often impose
constraints on everything from allocation size limits to the ways in which storage is
managed. Enterprise customers will also have to adjust to two major storage differences:
ephemeral storage and lack of shared storage.
• Perhaps one of the most disorienting features in the cloud is the use of ephemeral storage,
which means that if you turn off the server, or it has a hard fault, everything on the drive is
lost (data, boot parameters, updates, logs, etc.).
• This type of storage is fine for stateless servers (think web tier servers) which receive the
data they need from another source during operation, but is impossible to use for many
enterprise applications.
• The introduction of this type of storage into your operating environment adds a
management burden since you have to actively avoid using it for things that are important
to you.
88
89. 1. Rebuilding the Application Stack: Storage
• The second major storage difference is the general lack of shared storage in the cloud.
Shared storage is widely used in high availability and redundancy configurations, where if
one server goes down, others pick up the workload because they
• map to the same disk.
• Today’s clouds are unable to map a storage device to more than one server, so shared
storage in the cloud is currently not possible. As a result, high availability must be
achieved through some different and less proven architecture.
• This type of fundamental change highlights a major problem when adapting existing
applications to meet cloud requirements: the need to redesign the application to run
without a “tried and true” solution.
• Further, if the application is developed using third party software (such as Oracle), there
may be no opportunity to “redesign” it. Rather, you would have to select a different
product or manufacturer to get the necessary functionality.
89
90. 1. Rebuilding the Application Stack: Replicating Data
Centers
• Most enterprise applications work with a range of tools and services such as identity
management, monitoring, and directory services. When applications which rely on
these services are moved into the cloud, or new ones are created there, the
applications become disconnected from the data center, breaking important
relationships and dependencies.
• Therefore these key services and control processes need to be modified, replaced, or
possibly even eliminated to accommodate the cloud provider’s environment.
o Do you create a separate version of internal processes and control systems to run
independently within the cloud?
o Do you implement new services in the cloud with similar capabilities and find a way to tie them
back to the data center?
o Do you simply retool or build the application so it doesn’t depend on those services?
• The usual approach is to engage a professional services firm to assist in porting and
integrating the environment, or the cloud provider may provide similar services as
part of their managed hosting. The typical result is a lot of heavy (and expensive)
lifting in order to make it work.
90
91. SOA and the Cloud
Service Oriented Cloud Computing
Infrastructure
91
94. Outline: Part 3: Understanding the Components
• Summary: Web 2.0, SaaS, Utility Computing, Virtualization, SLAs, Autonomic computing, Grid
technology, Web Services, Service Oriented Architectures, Free and Open Source Software
• Deep Dive: Virtualization
o History
o Issues and Trends
o Supplier review: Citrix, IBM, Linux, Microsoft, Novell, Symantec, Oracle, VMWare
• Deep Dive: Data Centers
o History and the drive for efficiency and availability
o Changes and pressures on DC – drive for DC management
o Capex and opex DC costs
o DC economics drives cloud computing
• Deep Dive: Force.com, Google, Microsoft and Amazon
o Force.com
o Google App Engine
o Microsoft Azure
o Amazon Web Services
• Netflix deep dive
• AWS walk-through
94
95. Location and Scale: It’s Easier to
Ship Data than Power!
• Datacenters Are Popping Up in Surprising Places
o Quincy, WA
• Google, Microsoft, Yahoo!, and Others…
o San Antonio, TX
• Microsoft, US NSA, and Others…
Price per Kilo Where? Possible Reason Why
Watt Hour
3.6 cents Idaho Hydroelectric Power; Not Sent Long Distance
10.0 cents California Electricity Transmitted Long Distance over the Grid; Limited
Transmission Lines in the Bay Area; No Coal Fired Electricity
Allowed in California.
18.0 cents Hawaii Must Ship Fuel to Generate Electricity
95
96. Data Center Economics – simply scale
Scale is the main driver for cloud computing – its 5-7 times cheaper than doing it in house.
This is the fundamental principle of Amazon’s business model. So why focus on a private
cloud when it doesn’t have scale?
96
98. Mapping the Cloud Development Platform Landscape
Enterprise Centric
The challenge for Google and Amazon is can they
break out of the ‘geek developer’ into mainstream
enterprise, or will Oracle and IBM’s approach of
providing integrated web-centric and enterprise-
centric solutions be preferred by the buyers of
enterprise services?
See Oracle and IBM analysis for more details on their
approach.
Move into hosted /
Best managed solutions High
Effort Availability
Adding capabilities
Improving Availability
Web Centric 98
101. AWS Customers: Netflix.com
- More than 9 Billion USD market cap
- Migrating 100% on Amazon Web
Services
- So far: movie lists, website search,
transcoding, recommendations, etc.
- 24 M subscribers, 100k+ DVD titles
"AWS let us focus on innovation"
101
119. Outline: Part 4: Implementation
• Survey - what workloads others are moving into the cloud?
• Summary
o Key points in cloud migration
o Industry : Workload : Cloudability Space
• Project Plan – example from IBM
• Decision Tree for implementing Cloud Computing
o The Open Group decision tree
• Security
o Reviewing SAS70, PCI DSS, ISO27001, NIST, HIPAA, FISMA, CoBIT, Data
Protection Directive, practical aspects
• Architectural Review
• Concluding Remarks
119
123. Workloads ready for cloud computing
• Analytics • Desktop and devices
– Data mining, text mining or – Desktop
other analytics – Service/help desk
– Data warehouses or data marts • Development and test
– Transactional databases – Development environment
• Business services – Test environment
– Customer relationship • Infrastructure
management – Application servers
(CRM) or sales force automation – Application streaming
– E-mail – Business continuity/
– Enterprise resource planning disaster recovery
(ERP) applications – Data archiving
– Industry-specific applications – Data backup
• Collaboration – Data center network capacity
– Audio/video/Web conferencing – Security
– Unified communications – Servers
– VoIP infrastructure – Storage
– Training infrastructure
– Wide area network (WAN)
capacity
Source: IBM Market Insights, Cloud Computing Research, July 2009.
132. Elasticity, Risk, and User Incentives
Services Will Prefer Utility Computing to a Private Cloud When:
Demand Varies over Time Demand Unknown in Advance
Provisioning for Peak Leads to Web Startup May Experience a
Underutilization at Other Times Huge Spike If It Becomes Popular
Pay by the Hour Pay as You Go Does Not Require
(Even if the Hourly Rate is Higher) Commitment in Advance
The Value of Cost Associativity
UserHourscloud × (revenue – Costcloud) ≥
UserHoursdatacenter × (revenue – Costdatacenter )
Utilization
132
133. Cloud Is Mostly Driven by Money
Economics of Cloud Computing Are
Very Attractive to Some Users
Cloud Computing Will
Predicting Application
Track Cost Changes
Growth Hard
Better than In-House
Investment Risks May In-House, You Must
Be Reduced Provision for Peak
133
134. Cloud’s goal: flip this equation
30% 70%
On-Premise Your Managing All of the
Infrastructure Business “Undifferentiated Heavy Lifting”
Configuring
Cloud-Based More Time to Focus on
Your Cloud
Infrastructure Your Business
Assets
70% 30%
134
135. Benefits
• Shorter provisioning times
• Reduced capital outlay
• Allows more use of “try before you buy”
• Reduces the cost of “one-off” activities
• Costs associated with testing can be reduced
• Reduction in internal data center capacity
• Better architecture and design
• Consolidation and central administration:
• Greener IT
• Resources
• Improved administration and maintenance:
• Better quality services available from Cloud Computing:
• Better security
• Flexibility
• Improved financial control
135
136. Issues
• IT security and compliance
• Not mature
• Lack of clear definition of components
• Software licensing
• Service delivery clarity
• Calculating costs of service
• Integration
• Green IT
136
138. IBM Cloud Business Model
ROI Analysis Impact:
Reduction of Total Cost of Ownership of
Data Center Infrastructure
New
100% Development Liberated Reduced Capital Expenditure
funding for - Improved utilization reduces requirement for
Software new new capital purchases
Costs development, Strategic
transformatio Change Reduced Operations Expenditure
n investment Capacity - Lower facilities, maintenance, energy, IT
Power or direct service delivery and labor costs
Costs
saving
Additional Benefits
Deployment (1- - Reduced risk, less idle time, more efficient
Curren time) use of energy, acceleration of innovation
t IT Labor Costs
projects, enhanced customer service
Spend (Operations Software
and Costs
Maintenance)
Business Case Results
Power Costs Hardware,
labor &
Annual savings: $3.3M (84%)
(88.8%)
power from $3.9M to $0.6M
Hardware Labor Costs savings
Costs ( - 80.7%) reduced
Payback Period: 73 days
(annualized) annual cost
Hardware Costs of operation Net Present Value (NPV): $7.5M
( - 88.7%) by 83.8% Internal Rate of Return (IRR): 496%
Note: 3-Year Depreciation Period with 10% Discount Return On Investment (ROI): 1039%
Rate
139. Workloads ready for cloud computing
• Analytics • Desktop and devices
– Data mining, text mining or – Desktop
other analytics – Service/help desk
– Data warehouses or data marts • Development and test
– Transactional databases – Development environment
• Business services – Test environment
– Customer relationship • Infrastructure
management – Application servers
(CRM) or sales force automation – Application streaming
– E-mail – Business continuity/
– Enterprise resource planning disaster recovery
(ERP) applications – Data archiving
– Industry-specific applications – Data backup
• Collaboration – Data center network capacity
– Audio/video/Web conferencing – Security
– Unified communications – Servers
– VoIP infrastructure – Storage
– Training infrastructure
– Wide area network (WAN)
capacity
Source: IBM Market Insights, Cloud Computing Research, July 2009.
142. Source of Cloud Migration Delay and Blocking
• 1. Rebuilding the application stack within the cloud
• 2. Setting up the network
• 3. Adding end-to-end security
• 4. Managing the application in a separate environment
142