Windows AzureUma Plataforma para o Desenvolvimento de Aplicações

Pedro Rosa

http://netponto.org11ª Reunião Presencial - 19/06/2010

Pedro Rosa - DeepDiveIn

Contacto+351 961 624 203

LinkedInhttp://pt.linkedin.com/in/pedrobarraurosa

Twitterhttp://twitter.com/pedrorosa

Agenda

• Why Cloud?• What is it anyway?• How can I benefit?

Business Drivers for IT Projects

• Increase market share and revenue– Investing in product development and customer

facing interaction channels• Increase efficiency and lower TCO

– Investing in technologies and processes to drive efficiency and lower cost through optimization

Economy of ScaleLow High

ControlHigh Low

Control vs. Economy of Scale

Economy of ScaleLow High

ControlHigh Low

Build vs. Buy

This is not new…

Economy of ScaleLow High

ControlHigh Low

On premises vs. in the cloud

This is new…

Leveraging Economy of Scale

Ratio between $ buckets for medium and very large DC

Technology RatioNetwork 7.1Storage 5.7Administration 7.1

Source: Above the Clouds: A Berkeley View of Cloud Computing

http://netponto.org/

Short distance access to hydroelectric power

Long distance transmission over the grid

Energy generation based on expensive resources

Location Matters

Short distance access to hydroelectric power

Long distance transmission over the grid

Energy generation based on expensive resources

Price per KWH 3.6¢ Idaho 10.0¢ California

18.8¢ Hawaii

Location Matters

Application runs on-premises

Buy my own hardware, and

manage my own data center

Application runs at a hoster

Pay someone to host my

application using hardware that I

specify

Application runs using cloud

platform

Pay someone for a pool of computing resources that can be applied to a set

of applications

Datacenter Options

Datacenter Options

Application runs on-premises

• Bring my own machines, connectivity, software, etc.

• Complete control and responsibility

• Upfront capital costs for the infrastructure

Application runs at a hoster

• Rent machines, connectivity, software

• Less control, but fewer responsibilities

• Lower capital costs, but pay for fixed capacity, even if idle

Application runs using cloud

platform

• Shared, multi-tenant environment

• Offers pool of computing resources, abstracted from infrastructure

• Pay as you go

Software as a Service• Use applications over a network

Platform as a Service• Deploy customer-created applications to

a cloud

Infrastructure as a Service• Rent processing, storage, network

capacity, and other fundamental computing resources

X as a Service

IaaS

• User gets access to a pool of resources through VM abstraction

• User needs to patch/maintain the system• Scales only if stateless or if well partitioned

• E.g. Amazon EC2

PaaS

• User gets access to a pool of resources through a programming model

• System is managed by the cloud provider• Platform (programming model) can provide

scalability

• E.g. Windows Azure, Google App Engine

SaaS

• User gets access to applications• System is completely managed by the cloud

provider• Cloud provider may offer customization and

extensibility options

• E.g. Salesforce.com, Microsoft BPOS, Google Apps

http://netponto.org/

Need for Elasticity

Wall Street firm on Amazon EC2

3000 CPU’s for one firm’s risk management processes

Num

ber

of E

C2

Inst

ance

s

300 CPU’s on weekends

Thursday4/23/2009

Friday4/24/2009

Sunday4/26/2009

Monday4/27/2009

Tuesday4/28/2009

Saturday4/25/2009

Wednesday4/22/2009

3000--

300 --

Scale through PaaS

• The platform abstracts the physical infrastructure• Ability to scale built into the programming model

– Google’s AppEngine– Microsoft’s Windows Azure– Amazon’s elastic MapReduce– …

What is Windows Azure?An Operating System for the cloud

• Hardware Abstraction across multiple servers• Distributed Scalable, Available Storage• Deployment, Monitoring and Maintenance• Automated Service Management, Load Balancers,

DNS• Programming Environments• Interoperability• Designed for Utility Computing

Why Windows Azure?

• OS Takes care of your service in the cloud– Deployment– Availability– Patching– Hardware Configuration

• You worry about writing the service

What is Windows Azure?Features

• Automated Service Management• Compute• Storage• Developer Experience

What is Windows Azure?

Compute Storage

DeveloperSDK

DeveloperTools

What is Windows Azure?

Compute• .NET 4.0• Server 2008 – 64bit• Full Trust*• Web Role

• IIS7 Web Sites (ASP.NET, FastCGI)

• Web Services (WCF)

• Worker Role• Stateless Servers• Http(s), TCP

Storage

DeveloperTools

What is Windows Azure?

Compute

Storage• Durable, scalable, available• Blobs• Drives• Tables• Queues• REST interfaces

– Can be used without compute

All of the hardwareHardware Load BalancersServersNetworksDNSMonitoring

Automated service management

What is Windows Azure?

Compute Storage

DeveloperTools

What is Windows Azure?

Compute Storage

Developer SDK

• Windows Azure SDK– Local Development Fabric emulates running multiple

instances, load-balancing, service API– Local Development Storage emulates blob, tables and

queue storage– Command line tools allows integration outside of Visual

Studio

• Microsoft Visual Studio 2008, 2010 add-in– Provide rich tooling and templates to easily build your

application

Windows Azure API

• All operations are accessed through RoleEnvironment• RoleEnvironment Handles

– Retrieving configuration information– LocalResource locations (local disk)– Accessing Role information like InstanceEndpoints necessary for

inter-role communication– Exposes events during instance lifecycle, allowing a chance to

respond to topology or config change

Service Models

• Describes your Service<?xml version="1.0" encoding="utf-8"?><ServiceDefinition name="CloudService1" xmlns="http://schemas.microsoft.com/ServiceHosting/2008/10/ServiceDefinition"> <WebRole name="WebRole"> <ConfigurationSettings> <Setting name="AccountName"/> </ConfigurationSettings> <LocalStorage name="scratch" sizeInMB="50"/> <InputEndpoints> <!-- Must use port 80 for http and port 443 for https when running in the cloud --> <InputEndpoint name="HttpIn" protocol="http" port="80" /> </InputEndpoints> </WebRole> <WorkerRole name="WorkerRole"> <ConfigurationSettings> <Setting name="AccountName"/> <Setting name="TableStorageEndpoint"/> </ConfigurationSettings> </WorkerRole></ServiceDefinition>

Service Architecture

Service Architecture

My First Service (WorkerRole)

demo

Storage

• Blobs, Drives, Tables, Queues• Designed for the cloud

– 3 replicas– Guaranteed consistency

• Accessible directly from the internet via REST API – .NET Client library supported

• Does not require compute• Storage account drives unique URL, e.g.:

– https://<youraccount>.blob.core.windows.net

Blobs

• Blobs stored in Containers• 1 or more Containers per account

– Scoping is at container level– …/Container/blobname– $root is special name for root container

• Blobs– Two types, Page and Block

• Page is random R/W, Block has block semantics– Metadata, accessed independently

• name/value pairs (8kb total)

• Private or Public container access

Queues

• Simple asynchronous dispatch queue– Create and delete queues

• Message:– Retrieved at least once– Max size 8kb– Operations:

• Enqueue• Dequeue• RemoveMessage

Tables

• Entities and properties (rows & columns)• Tables scoped by account• Designed for billions+• Scale-out using partitions

– Partition key & row key– Operations performed on partitions– Efficient queries– No limit on number of partitions

• Use ADO.NET Data Services

Deploying Services

demo

Service Lifecycle

• Create service packageBinaries + Content + Service Metadata

• Deploy via web portal or Service Management API• Add & remove capacity via web portal or API• Deployed across fault domains• Upgrade with zero downtime

Automated Service Management

• You tell us what, we take care of how• What

– Service metadata• How

– Metadata describes service– No OS footprint– Service is copied to instances– Instances were copied to physical hardware– Physical hardware booted from VHD– All patching is performed offline

Service Monitoring

• SDK component providing distributed monitoring & data collection for cloud apps

• Support Standard Diagnostics APIs– Trace, Debug normally

• Manage multiple role instances centrally• Scalable• Choose what to collect & when to collect it

– Event Logs, Trace/Debug, Performance Counters, IIS Logs, Crash Dumps, Arbitrary log files

Design Considerations

• Scale and availability are the design points• Storage isn’t a relational database• Stateless

– Stateless front ends, store state in storage• Use queues to decouple components• Instrument your application (Trace)• Once you are on - stay on• Think about patching & updates

The CxO Value Proposition

• Time to Market– No upfront CAPEX– Instant access to compute resources

• Total Cost of Ownership – CAPEX/OPEX– Economy of scale– Elasticity

• Productivity– Higher level of abstraction for compute resources

• Risk Mitigation– Allows for trial and error– Unpredictable utilization patterns

Cloud Pricing

• Pricing Options– Pay as you go – flexible but unpredictable– Flat fees, Quotas, Limits allow for better control– Decisions based on business model, predictability, IT budgeting

• Pricing Dimensions– Compute pricing

• # CPUs (hours deployed vs. hours utilized)• Dedicated local storage & memory

– Bandwidth (inbound, outbound, inter-DC)• Can become very pricy for large storages• Disk shipments

– Storage• Average used storage capacity vs. , “fixe sized” storage• Storage transactions (read, write)• Used query time

The Bandwidth Challenge

• Storage vs. Bandwidth• Time and cost of transferring data

– AWS Import/Export• Compute and storage location matters

– Compute where the data is stored instead of moving the data to the compute resource

The Regulation Challenge

• Safe Harbor Act– Datacenter locations matter

• Patriot Act– Big Brother might get access to your data

• Industry Regulations– E.g. Payment Card Industry Data Security

Standard (PCI DSS)

Where to Start?

• Find a business case first!• Compute Scenarios

– Capabilities with little integration requirements• Input compute output

– High scalability/performance requirements– Challenging utilization requirements

• Storage Scenarios– Archiving solutions– Cross-corporate data sharing– Look for capabilities with infrequent compute

Questões?

Patrocinadores desta reunião

Próximas reuniões presenciais

• 19/06/2010 - Junho• 10/07/2010 - Julho• 14/08/2010 - Agosto• 18/09/2010 - Setembro

Reserva estes dias na agenda! :)

Obrigado!

Pedro Rosahttp://pt.linkedin.com/in/pedrobarraurosahttp://twitter.com/pedrorosa