Chris J.T. Auld Director, Intergen Limited New Zealand (Go All Blacks!)

Notes Before We Start Assumptions Basic understanding of application architecture Scale up vs Scale out Data partitioning/sharding No prior experience with Azure or any other cloud platform I Wear Yellow Crocs I work for Intergen I don’t work for Microsoft Contact me Chris Auld

A few slides on the Cloud value prop.

Intelligent Network Load Balancer Stateless ‘Worker’ Machines Shared Filesystem M Tiers of N stateless machines w/ Forward Caching Partitioned Relational DB Async Dispatch Queue Async Dispatch Queue ‘Monolithic’ State Tier Queue Activation Network Activation

Consolidation Cloud Consolidate traditional IT Move servers and apps out-house Reliable hardware Virtualized ‘traditional’ operating systems Self managed Scale Out Cloud Manage explosive growth Marginal costs count for everything ‘Disposable’ hardware Reliability through software High level management only

Power Usage Effectiveness = Total Facility Power / IT Equipment Power Running A Data Center is a Sisyphean Task You heat the air up…Only to cool the air back down again… PUE is basically a measure of how much ‘useful’ output you get. Anyone know the PUE of their data center? Intergen Wellington Data Center PUE is 1.62 Google and Microsoft are getting 1.10 to 1.25

Windows Azure Applications.NET Services Live Services SQL Azure Applications Others Windows Mobile Windows Vista/XP Windows Server The Azure Services Plaform An illustration

.NET Services Windows Azure Live Services Applications SQL Services Others Windows Mobile Windows Vista/XP Windows Server Fabric Storage Config Compute Application Windows Azure Windows in the cloud

Windows Azure Basics Windows Azure can potentially provide various kinds of Windows-based environments The current CTP release supports both.NET and unmanaged applications Windows Azure provides Operating System like concepts Process Management Computation Storage

… Fabric Compute Storage Application VMs Windows Azure Fabric Worker Role Instance Agent Windows Azure Compute Service A closer look Load Balancer HTTP IIS Web Role Instance

Windows Azure Compute Service Points of interest The VMs are provided by a cloud-optimized hypervisor Not your average VM They run 64-bit Windows Server 2008 Each VM has a one-to-one relationship with a processor core For developers: It’s mostly standard Windows and.NET A few things require accessing the Windows Azure Agent, e.g., logging A desktop facsimile of Windows Azure in the cloud is provided for development via the SDK

… Fabric Compute Storage Application Windows Azure Storage Service A closer look Blobs HTTP/ HTTPS TablesQueues

Windows Azure Storage Points of interest Storage types: Blobs: a simple hierarchy of binary data Tables: entity storage (not relational tables) Queues: high performance quasi-transactional queue structures Access: Data is exposed via a RESTful interface Data can be accessed by: Windows Azure applications Any other internet connected application on-premises or other cloud applications

Table... Table Entity... Entity Property... Property Azure Table Storage - Logical A closer look at tables Name Type Value Flexible, non schematized

Products... Orders People User... User Friend NameGender... DoB What Does Flexible Mean? DoB Date 1-Jan String

Azure Table Storage - Physical How Windows Azure Stores and Scales Your Data People PartKey: A RowKey: Name: David Anderson PartKey: B RowKey: Name: Simon Bruce PartKey: M RowKey: Name: Fred Matfield AB... U, V, W, X, Y, Z M

Windows Azure Storage Tables: Challenges Access via REST You can’t use ordinary ADO.NET SDK Provides some tooling No SQL No real joins, aggregates, etc. Very limited indexing An unfamiliar hierarchical structure You can’t easily move relational data to it Supporting services are scarce, e.g., reporting No schema

Windows Azure Storage Tables: Strengths Massive scalability By seamlessly providing scale-out data Applied to the right problem, Windows Azure Storage is a beautiful thing Amazon, Google, and others provide quite similar technologies This is approach is the state of the art for scale-out data

Web Role Instance Queue 1) Receive work 3) Dequeue message 4) Do work 2) Enqueue message 5) Delete message Worker Role Instance Windows Azure Storage: Queues The Asynchronous Dispatch Model

Fabric Controller Fabric Agent Storage Fabric Agent Web Role Instance Worker Role Instance The Windows Azure Fabric An illustration

Using Windows Azure Some examples A company might have a very elastic computing need 10 instances 24/7/ instances for 90 minutes 5 times per year A department might want agility without involving IT ‘Free’ to bypass usual organizational procedures

Using Windows Azure Some examples A start-up might create a new Web application on Windows Azure They can fail fast or scale fast An ISV might create a SaaS version of an existing.NET application on Windows Azure It’s.NET, so porting the code is doable An enterprise might build a new application on Windows Azure It could be a ‘short lived’ project

A lap around what others are doing

Vertically IntegratedPlatform Niche Player Pay As You Go Buy Up-front Microsoft Google Salesforce Microsoft Google Salesforce Amazon Mosso Amazon Mosso VMWare Appistry VMWare Appistry Trad Vendor v.Future Trad Vendor v.Future

VM EC2 Alternative Cloud Platforms Amazon Web Services Elastic Compute Cloud (EC2) provides VMs that can run Linux or Windows Linux or Windows Linux or Windows Windows Azure Fabric VM Windows Server 2008 Windows Azure Storage Windows Azure CTP Your Application Your Storage Your Application

Alternative Platforms Other Amazon Web Services technologies Amazon Web ServicesWindows Azure Blob storage Simple Storage Service (S3) Elastic Block Store Windows Azure Storage Blobs Scale-out storageSimpleDB Windows Azure Storage Tables Queues Simple Queue Service (SQS) Windows Azure Storage Queues

Alternative Platforms Google AppEngine Supports Java and Python Web applications Provides non-relational, scale-out storage Google AppEngine Windows Azure Fabric Windows Server 2008 Windows Azure Storage Windows Azure CTP Worker Role Web Role Datastore GQL Python/Java Runtime Web Application

Alternative Platforms Salesforce.com Force Platform A platform for data-driven enterprise applications Uses Apex, a Salesforce.com-defined language Provides non-relational, scale-out storage Force Database Force Runtime SOQLEnterprise Application

SQL Services.NET Services Windows Azure Live Services Applications Others Windows Mobile Windows Mobile Windows Vista/XP Windows Vista/XP Windows Server Windows Server SQL Azure A relational DB in the cloud SQL Azure Database Others (Future) “Huron” Data Hub

SQL Azure Today: SQL Azure Database Formerly known as SQL Server Data Services (SSDS) “Huron” Data Hub Built on the Microsoft Sync Framework In the future: Reporting Analysis Extract/Transform/Load (ETL) services More

Tabular Data Stream Database SQL Data Services “Huron” Data Hub Others (Future) SQL Data Services An illustration

Database Application Database SQL Data Services SQL Data Services Using one or multiple databases

SQL Data Services SQL Server Compact Edition Other Databases Data Sync “Huron” Data Hub "Huron" Data Hub An illustration

Using SQL Azure Some examples A Windows Azure application might use SQL Data Services for its data A departmental app could use SQL Data Services rather than a local database For better reliability, availability and data backup An organization might make data available to both in-house and partner apps through SQL Data Services Such as a company with a far-flung dealer network An enterprise might sync distributed data with the “Huron” Data Hub

.NET Services Windows Azure Live Services Applications SQL Services Others Windows Mobile Windows Mobile Windows Vista/XP Windows Vista/XP Windows Server Windows Server Service Bus Access Control ? ?.NET Services Infrastructure in the cloud

The Access Control Service The problem: Different organizations identify users with tokens containing different claims Applications can be faced with a confusing mess The solution: The Access Control Service implements a security token service (STS) in the cloud It accepts one token and issues another The claims in the outgoing token can differ from those in the incoming token An administrator can define rules for how this claims transformation is done

Access Control Interactions Your Access Control Project (a hosted STS) Your Access Control Project (a hosted STS) Relying Party (Your App) Relying Party (Your App) 2. Send Claims 4. Send Token (output claims from 3) 5. Send Message w/token 0. Certificate exchange; periodically refreshed Requestor (Your Customer) Requestor (Your Customer) 1. Define access control rules for a customer 6.Claims checked in Relying Party 3. Map input claims to output claims based on access control rules

Service Bus The problem: Exposing internal applications on the Internet isn’t easy Network address translation (NAT) and firewalls get in the way The solution: Service Bus provides a cloud-based intermediary between clients and internal applications It also provides a service registry that clients can use to find the services they need

Access Control Service Bus Registry Endpoints Organization Y Organization X Application Service Bus 2) Discover endpoints 1) Register endpoints 3) Access application 4) Upgrade to direct connection

Using.NET Services Some examples An app that’s accessed over the Internet from different organizations might rely on Access Control to rationalize the identity information it receives And to do access control An enterprise might expose an internal application to its trading partners via Service Bus

Conclusions Cloud platforms are here Microsoft is placing a big bet with the Azure Services Platform Elasticity is the sweet spot for early adopters A new world is unfolding Prepare to be part of it Architect for scale out today

Sessions On-Demand & Community Resources for IT Professionals Resources for Developers Microsoft Certification & Training Resources Resources Required Slide Speakers, TechEd 2009 is not producing a DVD. Please announce that attendees can access session recordings at TechEd Online. Required Slide Speakers, TechEd 2009 is not producing a DVD. Please announce that attendees can access session recordings at TechEd Online.

Required Slide Complete a session evaluation and enter to win! 10 pairs of MP3 sunglasses to be won

© 2009 Microsoft Corporation. All rights reserved. Microsoft, Windows, Windows Vista and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.