Transcript Future of Cloud Computing

The Future of Cloud Computing
The Next Ten Years
4 Dr. Rao Mikkilineni, Kawa Objects, Inc.
Vijay Sarathy, Kawa Objects Inc.
Kumar Malavalli, Kawa Objects, Inc.
Agenda
Lessons from the Past
4Waterwheel to Power Grid
4Evolution of Telecom Networks
The Present
4Why the present is not sustainable
4What must happen for progress
The Future
4Clouds and On-Demand Computing
4What it means for innovation
© 2009 KawaObjects, Inc.
2
4 Lessons from the
Past
© 2009 KawaObjects, Inc.
Credit
© 2009 KawaObjects, Inc.
Lessons from the Past
Early Power Generation
-The Waterwheel
The Modern Power Grid
The Telecom
Network
© 2009 KawaObjects, Inc.
5
Lessons from the Past
Local Power Generation
Modern Power
Grids
Telecom Network
Control
Local Control
Centralized Supply
Global
Interoperability
Users
Local
Ubiquitous
Ubiquitous
High Capital/Fixed costs
Commodity with
Economies of
Scale
Commodity with
Economies of
Scale
Economics
Efficiency
Difficult to scale up or
down; unable to address
over capacity or under
capacity; Redundant
Expenditure; Inefficient
overall
Managed
Resources.
Greater
Achieved
Productivity and
Innovation and
Innovation fueled
efficiencies through
through efficiencies
full automation
Lesson: Enterprises can not afford 2-5% of revenue spending on IT
© 2009 KawaObjects, Inc.
What this means for Industry
Commoditization of Computing is going to
happen
Disproportionate amount spent on
technology rather than business
Existing Management Complexity is not
Sustainable
Current Large Technology Incumbents
will have to evolve
Universal Access = GDP growth
© 2009 KawaObjects, Inc.
7
4 The Present
© 2009 KawaObjects, Inc.
Resources are Limited
© 2009 KawaObjects, Inc.
Today’s Business Drivers for Cloud Computing
1
Unexpected Demands created by Web 2.0
Consumer and Internet Applications
Explosion of social networking has created
wildly fluctuating demand straining the ability of
IT infrastructure to scale
Web 2.0 Applications
2
Explosion of Digitized Records and Content
E-mail
Records
Electronic
Health Records
Where we store everything ?
How do we retrieve it reliably ?
How do we secure it ?
How do we delete it ?
Digital Records
3
Ever escalating cost of improving ROI and
lowering TCO in the Datacenter
Rising Datacenter
complexity
© 2009 KawaObjects, Inc.
SAN, NAS, Virtualization
HA/DR, Performance Optimization,
Security
Rising TCO / Lower ROI
10
The Cloud must help reset complexity; TCO is getting out of Control
Actual TCO
Expected TCO
Virtualization
TCO
Bandwidth
Inversion
Cloud IT
1
Management
Complexity
Server-Centric
Computing
3
2
Resource
Consolidation
?
SAN Features
like Windowless
Backup
Self-Contained
1970
Shared Storage
2000
Cloud IT
Virtualized Datacenter
2005
2010
Time
© 2009 KawaObjects, Inc.
11
Evolution Of Backup and Disaster Recovery
RPO
RPO
Server-Centric
Computing
Shared
Storage
Live Migration
of Servers
1
Cloud IT
Virtualization
2
3
?
Days/Weeks
1970
Hours
2000
What Next?
Minutes/ Seconds
2005
2010
Time
© 2009 KawaObjects, Inc.
12
Cloud Computing as it is defined today…
A XaaS stack…
Web-based Services
Software-as-a-Service
…enables
Economies of Scale
Application-Components-as-a-Service
Compute
Compute
Resources
Resources
Software-Platform-as-a-Service
Virtual-Infrastructure-as-a-Service
End-to-end
Dynamic Resourcing
Physical-Infrastructure-as-a-Service
…On-Demand
Computing
..provides a
standard interface
Cloud API
What Is Lacking?
Server
© 2009 KawaObjects, Inc.
Network
Storage
13
The State of Clouds Today and What is Missing
To establish Trust in Cloud
Computing, this has to be
addressed end-to-end; not
just for servers
Massive Scalable
Reliable
For servers
Available
For servers
End-to-end Performance
Optimization
The Cloud
End-to-end
Dynamic Resourcing
End-to-end Security
Service Mediation
No Management
Complexity
© 2009 KawaObjects, Inc.
Global
Interoperability
14
Existing Infrastructure Silos and Multiple Inefficiencies
Application Specific
Server IT
Availability
Application Specific
Performance
Server Virtualization
Cache
Management
Application Specific
NetworkAdministration
IT
Network Virtualization
Server
Administration
System
Network
Administration
IP
Networks
Management
Management
Server
Availability
Network
IP
Performance
Networks
Availability
Management
Volume
Management
Performance
Management
© 2009 KawaObjects, Inc.
Performance
Security
SAN
Networks
Volume
Management
Performance
Application Specific
Application Specific
Security
Storage
IT
Storage Management
Storage Virtualization
DAS
Storage
Administration
Storage
NAS
Cache
COTS
Management
Storage
Management
Availability
Performance
Volume
Management
Security
SAN
Security
Performance
Management
Tiered
Configuration
Vision For a Next Generation Cloud based on Telecom History
Service Developers
Dynamic Composition &
Management of distributed
Logical Resources
Services
Distributed Service Delivery Platform
Service Operators
Distributed Service Creation Platform
Service Users
Fabric
Management Fabric
Distributed Service Assurance Platform
Distributed Service Mediation (Latency Based Connection FCAPS)
Virtual Computing QOS &
Dynamic FCAPS Management
Network QOS &
Dynamic FCAPS Management
Distributed Storage QOS &
Dynamic FCAPS Management
Distributed Hypervisors
Network Virtualization
Storage Virtualization
Server
Resources
Server
Resources
IP
Server
Resources
Server
Resources
iSCSI
Multi-CPU & Multi-Core Servers
© 2009 KawaObjects, Inc.
Fiber Channel
Infiniband
Multi-Latency Networks
Infrastructure
16 Fabric
COTS Storage
COTS Storage
COTS Storage
COTS Storage
Multi-Latency Storage Elements
16
The IT struggle today is about…
• IT Management
• SANs
• Proprietary HW with
advanced features
• Complexity
• Human Latency
End-to-end
Visibility &
Control
Costs
(Infrastructure,
Management and
Operations)
•
•
•
•
Clouds
Virtualization
Commodity HW
Simplicity
...balancing need for visibility & control with cloud scale & economics
© 2009 KawaObjects, Inc.
17
4 What the Future
Holds ..given past
experiences
© 2009 KawaObjects, Inc.
Cloud Computing Predictions for the next 10 Years
Expected TCO
End-to-end
Visibility
Co-existing
Private &
Public Clouds
CPU / Memory
Virtualization
Globally
interoperable,
massively scalable
service assurance,
creation, & delivery
platforms
Storage
Dynamic
Provisioning
TCO
1
End-to-end
Control
XaaS
in the cloud
End-to-end
Connection
FCAPS
Optimization
2
3
CPU to Spindle
Connection
FCAPS
Management
Dynamic Storage
capacity, Bandwidth and
Throughput Tuning
No Application SLA
2009
Storage SLA
2012
4
Workflow / Service SLA
CPU / Memory SLA
2015
2018
Time
© 2009 KawaObjects, Inc.
Customer Quote
“
I don’t care who provides my infrastructure even for my
mission critical applications as long as I have visibility into the
cloud and have control of my application response time, I/O,
throughput, availability, latency, security …and I have the
“
ability to adjust it based on my business priority and changing
workloads
- IT Business Alignment Manager at a Large Energy Company that is actively re-architecting their
datacenter around clouds
© 2009 KawaObjects, Inc.
20
What does Enabling End-to-End Visibility Mean?
Application-Centric Visibility
Virtual Appliances
App1
App 2
App 3
App 4
App 5
App 6
OS
OS
OS
OS
OS
OS
Hypervisor
Hypervisor
Servers
Server 1
Network
SAN
Server 2
NAS
IP
Storage
Storage 1
© 2009 KawaObjects, Inc.
Storage 2
21
What does Enabling End-to-End Control Mean?
Need to dynamically match
Application/Business
requirements to resources
App
B
App
A
Every app has different
resource needs
4
Latency
3
Throughput
Data
Pipe
2
IOPS ( IO operations/sec )
Resource
1
© 2009 KawaObjects, Inc.
22
Capacity
Coexistence of Private & Public Clouds
External
At Customer Site
Public Cloud 1
Services / Applications
Console
Server
A
2
1
2
3
Server
B
Storage
A
2
Kawa
DIME
Storage
B
Application /
Service Profile
Mediation
& Control
CPU
Cache
Storage
IOPS
……….
……….
Public Cloud 2
Server
C
Private
Cloud
3
Server
D
Storage
C
3
Storage
D
Kawa
DIME
Server
E
© 2009 KawaObjects, Inc.
1
1
Server
F
23
Kawa
Storage
Business Profile
RPO
RTO
Business
Priority
……….
……….
Coexistence of Private & Public Clouds
External
At Customer Site
Public Cloud 1
Services / Applications
Console
Server
A
2
1
2
3
Server
B
Storage
A
2
Kawa
DIME
Storage
B
Application /
Service Profile
Mediation
& Control
CPU
Cache
Storage
IOPS
……….
……….
Public Cloud 2
Server
C
Private
Cloud
3
Server
D
Storage
C
3
Storage
D
Kawa
DIME
Server
E
© 2009 KawaObjects, Inc.
1
1
Server
F
24
Kawa
Storage
Business Profile
RPO
RTO
……….
……….
……….
……….
4 Conclusion
© 2009 KawaObjects, Inc.
Vision For a Next Generation Cloud based on Telecom History
Service
Developers
Service
Users
Business
Services
Dial Tone
Business
Services Fabric
Distributed Service Delivery Platform
Management
Distributed Service Assurance Platform
Services Fabric
Infrastructure
Distributed Service Mediation (Latency Based Connection FCAPS)
Services Fabric
Virtual Computing QOS &
Dynamic FCAPS Management
Distributed Hypervisors
© 2009 KawaObjects, Inc.
Network QOS &
Dynamic FCAPS Management
Network Virtualization
26
Service
Operators
Distributed Service Creation Platform
Management
Services Dial
Tone
Distributed Storage QOS &
Dynamic FCAPS Management
Storage Virtualization
Computing
Services
Dial Tone
The Need for Open Standards, Global interoperability and Massive Scaling
Power Distribution Grids
Standards/Regulatory Bodies
ANSI
IEEE
ASME
FERC
Scale: Billions of consumers
Telecom Networks
Standards Body
ITU
Scale: Billions of users
Standards Body
IETF
Internet
Scale: Billions of devices
Standards Body
Cloud
Computing
???
Applications
Scale: Trillions of services
© 2009 KawaObjects, Inc.
27
27
Thank You!
{ Questions }
4 [email protected]
[email protected]