Transcript
Introduction to Cloud Computing ....... for Enterprise Users
Lew Tucker, Ph.D.
CTO, Cloud Computing Sun Microsystems, Inc.
Cloud computing means different things to different people
… and covers a lot of territory
Software as a Service Utility Computing Platform as a Service Virtualization Grid Computing
Database as a Service Application Hosting Infrastructure as a Service
Storage as a Service
Cloud Computing
Traditional DataCenters
is it the future?
Cloud Computing Public or Private
Enterprise IT Web Companies
2000
2005
2010?
2015?
2020?
2025?
Alternative to traditional data centers
“Let me be very clear here: I really don’t want to operate datacenters anymore... We’d rather spend our time giving our customers great service and writing great software rather than managing physical hardware,”
Don MacAskill, CEO, Smugmug
Definition by NIST
Cloud computing is a model for on-demand network access to a shared pool of configurable computing resources that can be rapidly provisioned and released with minimal management effort or service provider interaction.
http://csrc.nist.gov/groups/SNS/cloud-computing/index.html
Cloud Computing Attributes
Pay per use Elastic Self Provisioning Through Software Simple Scalable Services Virtualized Physical Resources Highly Automated Operations
Benefits: Efficiency, Flexibility and Speed
Economics Developer Centric Flexibility
Pay As-You-Go Op-ex vs. Cap-ex Virtualization
Rapid, Self Provisioning Faster Deployment API-Driven
Highly Elastic On Demand Scalable Services
Number of players rapidly expanding
Cloud Computing 101
Cloud Computing Models
Software as a Service
Applications on-demand
Platform as a Service
Developer platform for creating applications
Infrastructure as a Service
Storage and compute capabilities offered as a service
Natural Evolution of the Web
Web Sites Applications
SaaS
Developer Platforms
PaaS
Compute and Storage
IaaS
Software as a Service (SaaS)
Applications on demand: • Subscription-based, multi-tenant, nothing to download or manage
• • • •
Google Apps (docs, email) Microsoft Exchange Online Yahoo Mail TurboTax Online
• • • •
Salesforce.com NetSuite Oracle CRM On-Demand Cisco WebEx Weboffice
Platform as a Service (PaaS)
On-demand develop and deploy apps • Unique programming model, auto-scaling • Often both a platform and a channel
• Google AppEngine • Force.com (salesforce.com) • Netsuite Business OS
• Heroku • Aptana Cloud Connect • Facebook
Infrastructure as a Service (IaaS)
On-demand virtual infrastructure • Lowest level, most general, self-provisioning • Unlimited managed resources
• Amazon AWS (EC2, S3, SQS) • Microsoft Azure • RackSpace Cloud
• Savis • Terremark • Joyent
Two more core concepts
• Virtual Machine Images
> Complete, pre-configured, image of application and OS > Pre-packaged or built by user
• Cloud APIs
> Programmic way to provision and manage compute,
storage, and network resources > Access to scalable services (S3, SimpleDB) > Work underway to standardize for interoperability
Amazon's AWS
3202 Machine Images
Issues – trusting a service provider
• Data governance and application security
> Who has access > Trust in security of a shared, multi-tenant environment
• Legal
> Who can see my data, where is it? > Third-party involvement in discovery > Regulatory compliance
• Business
> Reliability > Lock-in
Public vs Private Clouds
Public Private Hybrid
Pay as you go, multitenant applications and services Access virtually unlimited resources
Cloud Computing model in a company's own datacenter Resources directly owned but therefore constrained
Mixed usage of both public and private clouds, often integrated into the same application
Emerging usage of both public and private clouds in the enterprise
Public Cloud (service)
• Rapid provisioning of almost unlimited resources • Pay only for what you need • Opex vs Capex • Departmental projects • Analytics, Dev/test • Customer-facing apps
Private Cloud (on-premise)
• Efficiencies of virtualization and data center automation • Dynamic re-allocation of resources • Reduction in operating costs • Departmental self-service and chargeback
Software and services vendors for building private clouds
• • • • • • • • VMware Citrix Eucalyptus Appistry Univa 3Tera Sun, Oracle, IBM, HP, Cisco Accenture, Deloite
IT-built cloud for internal customers using a shared pool of virtual resources
App App App App OS + Virtualization OS Hardware
OS + Virtualization Hardware
App
App
App
App OS + Virtualization OS Hardware
App
App
App
App
OS + Virtualization Hardware
OS + Virtualization Hardware
Department A
Department B
Department C
Expanded Role for IT
• Deliver the best technology solution for the business balancing cost, security, speed, user experience
> Greater agility for business units through more options and self-
service
• IT looks more like a business partner and service provider
> Higher degree of automated system administration > Outsource many functions to pubic cloud providers
• Responsible for IT policy, security, best practices
Cloud Usage Patterns
Test and Development Functional Offload
(Batch Processes – TimesMachine)
Functional Offload
(Storage – SmugMug)
(Temporary Peak Load )
Cloud Bursting Web Service
GSA Cloud Computing Storefront
IaaS Providers IaaSProviders
IaaS Vendor IaaSVendor 1
Government Agencies
Once IaaS Services are procured the Federal Agency works directly with the selected IaaS vendor in configuring and utilizing the services via the Internet
IaaS Vendor 2 IaaS Vendor n
Federal Agency 1 Federal Agency 2 Federal Agency n
Internet
4
3 Based on Federal Agency’s selection, the GSA Cloud Storefront enables the procurement of IaaS services with the vendor.
1
GSA Cloud Storefront (Web Portal)
2
Federal Agencies inquire and procure IaaS service through the GSA Cloud Storefront
The GSA Federal Cloud Storefront provides the predefined IaaS service offering options from the supported IaaS vendors based on the submitted inquires from the Federal Agency
Public Cloud
USA.gov and Data.gov
• • • • Federal government's most high-profile websites Reduction in annual expenses $2.5M to $800k Deploy new apps in 24 hours Built on Terremark's Enterprise Cloud platform
Public Cloud
MedCommons: health records sharing
• Health records services provider allowing consumers to share info using Amazon's AWS • Leverages Amazon's billing systems • Subscription-based • HIPPA compliant
http://aws.amazon.com/solutions/case-studies/medcommons/
Private Cloud
Washington Mutual: internal cloud
• Goal: reduce waste, complexity, and costs • Multi-phase approach
> > > >
Started with existing physical servers (< 10% utilization) Moved to a fixed number of virtual servers (VMware) Dynamic allocation of VMs (30-day increments) Self-service, compute on demand, highly automated ops
• Experience
> Utility computing 40-70% more cost effective > Unit costs of individual VMs down 60% in 18 months > Deployment now < 5 days
Private Cloud
Washington Mutual - advice
• Standardize and simplify all offerings • Use server consolidation to drive critical mass • Automate and deploy tools to drive transparency for platform users “I see the difference between utility and cloud-based computing is that the enterprise cloud really focuses on true, on-demand compute. When somebody needs it, they get it. And if they need it for one day, they get it for just one day” - Barton Warner, VP
Internal clouds as the next generation of data center management
• Builds on virtualization and server consolidation • Improves cost-effectiveness of IT workloads • Highly automated, removes operators from provisioning • Provides departmental self-service • Improves business agility up to the constraint of fixed resources (is this still a cloud?) • Potentially capable of “bursting” out to public clouds
New Data and Management Economics Compute Trend
New Analytics Emerge
(MapReduce, Hadoop...)
Data (Storage) Trend
Semi-structured Data
(Mogile, Bigtable, HDFS...)
Master/Slave
Architectural shift to the cloud and HPC-style workloads Open source, general purpose datawarehouse
Object Store
Semi-structured Database
ScaleDB, Big Table, SimpleDB hBase Master/Master
Proprietary, dedicated datawarehouse
OLTP is the datawarehouse
Distributed FS
Unstructured Data
Federated/ Sharded
Structured Data
Hybrid Cloud
GeoEye: satellite and aerial imagery
• Satellite and aerial imagery and geospatial products for national security and commercial customers • Combine on-premise Appistry cloud with Amazon's S3 service • Appistry CloudIQ uses 50+ servers for image processing • Amazon's S3 serves up imagery to customers
http://www.appistry.com/customers/profile/geoeye
Agile Analytics in the Cloud
Enterprise software vendors team up to provide a business intelligence solution that any business can setup in a matter of minutes.
It's Not Just About Cheap Computing
Business Agility
Web Scale
THE NETWORK
is YOUR Computer
Best way is to simply try it yourself
Cloud Computing Resources
• Sun Resources
> > > > > > > > > > >
https://www.sun.com/offers/details/cloud_computing_primer.xml https://www.sun.com/offers/details/CloudComputing.xml http://www.sun.com/solutions/cloudcomputing/perspectives.jsp http://kenai.com/projects/suncloudapis http://developers.sun.com/cloud/ http://blogs.sun.com/ec2/entry/hardened_opensolaris_2008_11_on http://kenai.com/projects/s3-crypto/pages/Home http://kenai.com/projects/zfs-backup-to-s3/pages/Home http://www.sun.com/service/cloud/ http://wikis.sun.com/display/cloud/Patterns http://wikis.sun.com/display/VeriScale/Home
• Other Resources
> > > > > >
http://groups.google.com/group/cloud-computing?lnk= http://groups.google.com/group/cloudforum?lnk= http://searchcloudcomputing.techtarget.com/ http://cloudcomputing.sys-con.com/ http://cloudbook.net/ http://www.eucalyptus.com/