CLOUD COMPUTING – AN OVERVIEW

ReddyRaja
Research Consultant, IIIT Hyderabad
Vasudev Verma
Associated Professor, IIIT Hyderabad
 Cloud Computing - Some terms
 Term cloud is used as a metaphor for internet
 Concept generally incorporates combinations of the following
 Infrastructure as a service (IaaS)
 Platform as a service (PaaS)
 Software as a service(SaaS)
 Not to be confused with
 Grid Computing – a form of distributed computing
 Clusterof loosely coupled, networked computers acting in concert to
perform very large tasks
 Utility Computing – packaging of computing resources such as
computing power, storage, also a metered services
 Autonomic computing – self managed
 Grid Computing
 Share Computers and data
 Evolved to harness inexpensive computers in Data center to solve variety of problems
 Harness power of loosely coupled computers to solve a technical or mathematical problem
 Used in commercial applications for drug discovery, economic forecasting, sesimic analysis and back-office
 Small to big
 Can be confined to a corporation
 Large public collaboration across many companies and networks
 Most grid solutions are built on
 Computer Agents
 Resource Manager
 Scheduler
 Compute grids
 Batch up jobs
 Submit the job to the scheduler, specifiying requirements and SLA(specs) required for running the job
 Scheduler matches specs with available resources and schedules the job to be run
 Farms could be as large as 10K cpus
 Most financial firms has grids like this
 Grids lack automation, agility, simplicity and SLA guarantees
 Utility Computing
 More related to cloud computing
 Applications, storage, computing power and network
 Requires cloud like infrastructure
 Pay by the drink model
 Similar to electric service at home
 Pay for extra resources when needed
 To handle expected surge in demand
 Unanticipated surges in demand
 Better economics
 Cloud computing – History
 Evolved over a period of time
 Roots traced back to Application Service Providers
in the 1990’s
 Parallels to SaaS
 Evolved from Utility computing and is a broader
concept
 Cloud computing
 Much more broader concept
 Encompasses
 IIAS, PAAS, SAAS
 Dynamic provision of services/resource pools in a co-ordinated fashion
 On demand computing – No waiting period
 Location of resource is irrelevant
 May be relevant from performance(network latency) perspective, data locality
 Applications run somewhere on the cloud
 Web applications fulfill these for end user
 However, for application developers and IT
 Allows develop, deploy and run applications that can easily grow capacity(scalability), work
fast(performance), and offer good reliability
 Without concern for the nature and location of underlying infrastructure
 Activate, retire resources
 Dynamically update infrastructure elements without affecting the business
 Clouds Versus Grids

 Clouds and Grids are distinct

 Cloud
 Full private cluster is provisioned
 Individual user can only get a tiny fraction of the total resource pool
 No support for cloud federation except through the client interface
 Opaque with respect to resources
 Grid
 Built so that individual users can get most, if not all of the resources in
a single request
 Middleware approach takes federation as a first principle
 Resources are exposed, often as bare metal
 These differences mandate different architectures for each
 Cloud Mythologies

 Cloud computing infrastructure is just a web service interface to

operating system virtualization.
 “I’m running Xen in my data center – I’m running a private cloud.”
 Cloud computing imposes a significant performance penalty over
“bare metal” provisioning.
 “I won’t be able to run a private cloud because my users will not
tolerate the performance hit.”
 Clouds and Grids are equivalent
 “In the mid 1990s, the term grid was coined to describe technologies
that would allow consumers to obtain computing power on demand.”
Commercial clouds
 Cloud Anatomy

 Application Services(services on demand)

 Gmail, GoogleCalender
 Payroll, HR, CRM etc
 Sugarm CRM, IBM Lotus Live
 Platform Services (resources on demand)
 Middleware, Intergation, Messaging, Information, connectivity etc
 AWS, IBM Virtual images, Boomi, CastIron, Google Appengine
 Infrastructure as services(physical assets as services)
 IBM Blue house, VMWare, Amazon EC2, Microsoft Azure Platform, Sun Parascale and more
Cloud Computing - layers

Layers Architecture
What is a Cloud?

Individuals Corporations Non-Commercial

Cloud Middle Ware

Storage OS Network Service(apps) SLA(monitor),
Provisioning Provisioning Provisioning Provisioning Security, Billing,
Payment

Resources
Services Storage Network OS
 Why cloud computing
 Data centers are notoriously underutilized, often idle 85%
of the time
 Over provisioning
 Insufficient capacity planning and sizing
 Improper understanding of scalability requirements etc
 including thought leaders from Gartner, Forrester, and IDC
—agree that this new model offers significant advantages
for fast-paced startups, SMBs and enterprises alike.
 Cost effective solutions to key business demands
 Move workloads to improve efficiency
 How do they work?
 Public clouds are opaque
 What applications will work well in a cloud?
 Many of the advantages offered by Public Clouds appear useful for “on
premise” IT
 Self-service provisioning
 Legacy support
 Flexible resource allocation
 What extensions or modifications are required to support a wider variety of
services and applications?
 Data assimilation
 Multiplayer gaming
 Mobile devices
 Cloud computing - Characteristics
 Agility – On demand computing infrastructure
 Linearly scalable – challenge
 Reliability and fault tolerance
 Self healing – Hot backups, etc
 SLA driven – Policies on how quickly requests are processed
 Multi-tenancy – Several customers share infrastructure, without
compromising privacy and security of each of the customer’s data
 Service-oriented – compose applications out of loosely coupled services. One
service failure will not disrupt other services. Expose these services as API’s
 Virtualized – decoupled from underlying hardware. Multiple applications can
run in one computer
 Data, Data, Data
 Distributing, partitioning, security, and synchronization
Public, Private and Hybrid clouds
 Public clouds
 Open for use by general public
 Exist beyond firewall, fully hosted and managed by the vendor
 Individuals, corporations and others
 Amazon's Web Services and Google appEngine are examples
 Offers startups and SMB’s quick setup, scalability,
flexibility and automated management. Pay as you go
model helps startups to start small and go big
 Security and compliance?
 Reliability concerns hinder the adoption of cloud
 Amazon S3 services were down for 6 hours
 Public Clouds (Now)

 Large scale infrastructure available on a rental basis

 Operating System virtualization (e.g. Xen, kvm) provides CPU isolation
 “Roll-your-own” network provisioning provides network isolation
 Locally specific storage abstractions
 Fully customer self-service
 Service Level Agreements (SLAs) are advertized
 Requests are accepted and resources granted via web services
 Customers access resources remotely via the Internet
 Accountability is e-commerce based
 Web-based transaction
 “Pay-as-you-go” and flat-rate subscription
 Customer service, refunds, etc.
 Private Clouds
 Within the boundaries(firewall) of the organization
 All advantages of public cloud with one major difference
 Reduce operation costs
 Has to be managed by the enterprise
 Fine grained control over resources
 More secure as they are internal to org
 Schedule and reshuffle resources based on business demands
 Ideal for apps related to tight security and regulatory concerns
 Development requires hardware investments and in-house
expertise
 Cost could be prohibitive and cost might exceed public clouds
 Clouds and SOA

 SOA Enabled cloud computing to what is today

 Physical infrastructure like SOA must be discoverable, manageable and
governable
 REST Protocol widely used(Representational State Transfer)
 Clouds for Developers
 Ability to acquire, deploy, configure and host
environments
 Perform development unit testing, prototyping and
full product testing
 Open Source Cloud Infrastructure

 Simple
 Transparent => need to “see” into the cloud
 Scalable => complexity often limits scalability
 Secure => limits adoptability
 Extensible
 New application classes and service classes may require new features
 Clouds are new => need to extend while retaining useful features
 Commodity-based
 Must leverage extensive catalog of open source software offerings
 New, unstable, and unsupported infrastructure design is a barrier to uptake,
experimentation, and adoption
 Easy
 To install => system administration time is expensive
 To maintain => system administration time is really expensive
 Microsoft and Amazon face challenges

 Globus/Nimbus
 Client-side cloud-computing interface to Globus-enabled TeraPort cluster at U of C
 Based on GT4 and the Globus Virtual Workspace Service
 Shares upsides and downsides of Globus-based grid technologies
 Enomalism (now called ECP)
 Start-up company distributing open source
 REST APIs
 Reservoir
 European open cloud project
 Many layers of cloud services and tools
 Ambitious and wide-reaching but not yet accessible as an implementation
 Eucalyptus
 Cloud Computing on Clusters
 Amazon Web Services compatible
 Supports kvm and Xen
 Open Nebulous

 Joyent
 Based on Java Script and Git
 Open Source Cloud Ecosystem - Tools

 RightScale
 Startup focused on providing client tools as SaaS hosted
in AWS
 Uses the REST interface
 Canonical
 Ubuntu 9.10 (Karmic Koala)
 Includes KVM and Xen Hypervisors
 Open Source Cloud Anatomy

 Extensibility
 Simple architecture and open internal APIs
 Client-side interface
 Amazon’s AWS interface and functionality (familiar and testable)
 Networking
 Virtual private network per cloud
 Must function as an overlay => cannot supplant local networking
 Security
 Must be compatible with local security policies
 Packaging, installation, maintenance
 system administration staff is an important constituency for uptake
 Open Source Cloud Anatomy .. cntd

 Private clouds are really hybrid clouds

 Users want private clouds to export the same APIs as the public clouds
 In the Enterprise, the storage model is key
 Scalable “blob” storage doesn’t quite fit the notion of “data file.”
 Cloud Federation is a policy mediation problem
 No good way to translate SLAs in a cloud allocation chain
 “Cloud Bursting” will only work if SLAs are congruent
 Customer SLAs allow applications to consider cost as first-class
principle
 Buy the computational, network, and storage capabilities that are required
Open Source Clouds contd.
Eucalyptus (Elastic Utility Computing Architecture Linking Your Programs To
Useful Systems)
 Clouds and Virtualization

 Operating System virtualization (Xen, KVM, VMWare, HyperV) is only

apparent for IaaS
 AppEngine = BigTable
 Hypervisors virtualize CPU, Memory, and local device access as a single
virtual machine (VM)
 IaaS Cloud allocation is
 Set of VMs
 Set of storage resources
 Private network
 Allocation is atomic
 SLA
 Monitoring
 Cloud Performance

 Extensive performance study using HPC applications and

benchmarks
 Two questions:
 Performance impact of virtualization
 Performance impact of cloud infrastructure
 Observations:
 Random access disk is slower with Xen
 CPU bound can be faster with Xen -> depends on configuration
 Kernel version is far more important
 No statistically detectable overhead
 AWS small appears to throttle network bandwidth and (maybe) disk
bandwidth -> $0.10 / CPU hour
Cloud Infrastructure
 Network operations center

 Physical Infrastructure
Cloud Infrastructure ..contd
 Physical Security

 Cooling
Cloud Infrastructure ..contd
 Power infrastructure, Network Cabling, Fire
safety
Clouds – open for innovation
 Cloud computing open issues
 Governance
 Security, Privacy and control
 SLA guarantees
 Ownership and control
 Compliance and auditing
 Sarbanes and Oxley Act
 Reliability
 Good servive provider with 99.999% availability
 Cloud independence – Vendor lockin?
 Cloud provider goes out of business
 Data Security
 Cloud lockin and Loss of control
 Plan for moving data along with Cloud provider
 Cost?
 Simplicity?
 Tools
 Controls on sensitive data?
 Out of business
 Big and small
 Scalability and cost outweigh reliability for small businesses
 Big businesses may have a problem
 Cloud articles
 http://blogs.zdnet.com/Hinchcliffe/?p=488&tag=bt
xcsim
 http://blogs.zdnet.com/Howlett/?p=558&tag=btxcsi
m
 http://blogs.zdnet.com/BTL/?p=9560&tag=btxcsim
 http://www.businessweek.com/technology/content/
aug2008/tc2008082_445669_page_3.htm
 http://www.ibm.com/developerworks/websphere/te
chjournal/0904_amrhein/0904_amrhein.html
 http://cloudcomputing.sys-con.com/
 Battle in the cloud
 Amazon Web Services
 Google App Engine
 Free upto 500 MB,
 Freefor small scale applications?
 Universities?
 Pay when you scale
 GoGrid
 .. Some more Hosting companies
 Where is HP, IBM, Oracle(+sun) and Dell?