Cloud application architecture with sql azure and windows azure

Cloud Application Architecture with SQL Azure and Windows AzureDr. Eduardo Castro MartínezMicrosoft [email protected]

http://comunidadwindows.orghttp://ecastrom.blogspot.com

Session Objectives

Provide a big-picture view of the SQL Server in the Cloud

Provide a big-picture view of the Windows Azure platform

Describe typical scenarios for using the Windows Azure platform

Agenda

Cloud Computing PlatformUnderstanding Cloud ComputingCloud Computing Security Cloud Computing and Standards

Creating Cloud Aware ApplicationsWindows Azure Platform

4

Computing Evolution

Now40’s 70’s 80’s

•Mobility explosion•Collaboration•High level of interconexion between users•Web platform growing•Pressure over companies data centre.

http://images.google.es/imgres?imgurl=http://www.opentheiphone.org/OpenIPhoneLogo.png&imgrefurl=http://www.opentheiphone.org/mainsite/&usg=__D2DmtZaKl10xBExUGw5KDuF9H0Y=&h=400&w=210&sz=86&hl=es&start=13&tbnid=G3XgnJT28_ad6M:&tbnh=124&tbnw=65&prev=/images?q=iphone+logo&gbv=2&hl=es

Issues Facing Developers

70% of Web Development Effort is “Muck”: Data Centers Bandwidth / Power / Cooling Operations Staffing

Scaling is Difficult and Expensive: Large Up-Front Investment Invest Ahead of Demand Load is Unpredictable

Solution: Web-Scale ComputingScale capacity on demand

Turn fixed costs into variable costs

Always available

Rock-solid reliability

Simple APIs and conceptual models

Cost-effective

Reduced time to market

Focus on product & core competencies

THE CLOUD GENERAL CONCEPTS

Typical Application Grow?

Users

Servers

Add some Web Servers

Buy bigger SQL Box

Rearchitect Database to

scale out

Buy more of everything

Unusedcapacity

Workload Patterns Optimal For Cloud

Usage

Com

pu

te

Time

Average

Inactivity

Period

“On and Off “

On & off workloads (e.g. batch job)Over provisioned capacity is wasted Time to market can be cumbersome

Com

pu

te

Time

“Unpredictable Bursting“

Average Usage

Unexpected/unplanned peak in demand Sudden spike impacts performance Can’t over provision for extreme cases

Average Usage

Com

pu

te

Time

“Growing Fast“

Successful services needs to grow/scale Keeping up w/ growth is big IT challenge Complex lead time for deployment

Com

pu

te

Time

Average Usage

“Predictable Bursting“

Services with micro seasonality trends Peaks due to periodic increased demandIT complexity and wasted capacity

What is the ‘Cloud’?

The Cloud – metaphor for the internet‘Cloud Computing’Exploiting resources ‘out there’

Scalable computational powerData management and storageEnabling servicesApplications

On Premises Hosted VendorCloud

Bu

ild v

s. B

uy

Build

Buy “Packaged

”Application

An application that I buy “off the shelf” and

run myself

“Custom Built”

Application

An application that I develop and run myself

Hosted “Custom

Built”

An application that I develop myself and run at a host site

Hosted “Packaged

”

An application that I buy “off the shelf” and run at a host

site

“Custom Built”

using cloud services

An application that I develop

myself and host using

cloud platform

“Packaged”

Using cloud services

An application that I buy “off the shelf” and

host using cloud platform

“Software as a

Service”

A hosted application

that I buy as a service from a

vendor

“Platform as a

Service”

A vendor hosted

development and runtime environment

What is the ‘Cloud’?

The Cloud

3 Cloud Service Models

Cloud Software as a Service (SaaS)Use provider’s applications over a network

Cloud Platform as a Service (PaaS)Deploy customer-created applications to a cloud

Cloud Infrastructure as a Service (IaaS)Rent processing, storage, network capacity, and other fundamental computing resources

To be considered “cloud” they must be deployed on top of cloud infrastructure that has the key characteristics

12

Service Model Architectures

Cloud Infrastructure

IaaS

PaaS

SaaS

Infrastructure as a Service (IaaS) Architectures

Platform as a Service (PaaS)Architectures

Software as a Service (SaaS)

Architectures


SaaS


PaaS

SaaS


IaaS

PaaS


PaaS


IaaS

13

Cloud App Architecture

Intelligent Network Load Balancer

Stateless ‘Worker’ Machines

Shared Filesyste

m

M Tiers of N stateless machines w/ Forward Caching

Partitioned Relational

DB

Async DispatchQueue

‘Monolithic’State Tier

Queue Activation

Network Activation

High Scale Application Archetype

Cloud aware Architecture

SQL Azure

Common DB

Tables

Castellan Venue

DB

Castellan Venue DB’s

ShardingPartition

Castellan Venue

DB


ShardingPartition

Castellan Venue

DB


ShardingPartition

...

Windows Azure Roles

http:// gagame .* Dynamic Worker(tasks

uploaded as blobs and sharding)

Partitioner Worker

Windows Azure StorageQueues for communication between clients and roles

-- - --- - -

-- - --- - -

Tables to record server & partition information

Blobs to store web and worker role resources

Client Applications

C# Application on Premises

Silverlight Application• Common Educational Experiences

Service Bus

WCF

On PremisesSQL Server

Configuration Statistics

Distributed Cache Worker

Sample Cloud Application Architecture

Source: [email protected]

Sample Cloud Architecture

Source: [email protected]

19

The ‘Why’ and ‘How’ of Cloud Migration

There are many benefits that explain why to migrate to clouds

Cost savings, power savings, green savings, increased agility in software deployment

Cloud security issues may drive and define how we adopt and deploy cloud computing solutions

20

Balancing Threat Exposure and Cost Effectiveness

Private clouds may have less threat exposure than community clouds which have less threat exposure than public clouds. Massive public clouds may be more cost effective than large community clouds which may be more cost effective than small private clouds. Doesn’t strong security controls mean that I can adopt the most cost effective approach?

21

Cloud Migration and Cloud Security Architectures

Clouds typically have a single security architecture but have many customers with different demands

Clouds should attempt to provide configurable security mechanisms

Organizations have more control over the security architecture of private clouds followed by community and then public

This doesn’t say anything about actual security

Higher sensitivity data is likely to be processed on clouds where organizations have control over the security model

22

Putting it Together

Most clouds will require very strong security controlsAll models of cloud may be used for differing tradeoffs between threat exposure and efficiencyThere is no one “cloud”. There are many models and architectures.How does one choose?

23

Migration Paths for Cloud Adoption

Use public cloudsDevelop private clouds

Build a private cloudProcure an outsourced private cloudMigrate data centers to be private clouds (fully virtualized)

Build or procure community cloudsOrganization wide SaaSPaaS and IaaSDisaster recovery for private clouds

Use hybrid-cloud technologyWorkload portability between clouds

24

Possible Effects ofCloud Computing

Small enterprises use public SaaS and public clouds and minimize growth of data centersLarge enterprise data centers may evolve to act as private cloudsLarge enterprises may use hybrid cloud infrastructure software to leverage both internal and public cloudsPublic clouds may adopt standards in order to run workloads from competing hybrid cloud infrastructures

An Overview of the Windows Azure Platform

The Windows Azure Platform

Windows AzureSQL AzureWindows Azure platform AppFabricDeveloper tools

Windows Azure Basics

Windows Azure can potentially provide various kinds of Windows-based environments

The current release supports both .NET and unmanaged applications

Windows Azure provides Operating System like concepts

Process ManagementComputationStorage

Azure™ Services Platform

A set of connected serversOn which developers can:

Install and run servicesStore and retrieve data

What Is The Cloud in Windows Azure?

It is an operating system for the cloudIt is designed for utility computingIt has four primary features:

Service managementComputeStorageDeveloper experience

What Is Windows Azure?


SQL Azure

Windows Azure

CloudApplicatio

ns Windows Azure AppFabric

On-PremisesApplicatio

nsOthersWindows

…

Windows AzureAn illustration

Fabric

Compute Storage

Config

Application

SQL Azure

Windows Azure

Applications

Applications

OthersWindows

Windows Azure AppFabric

Windows Azure Datacenter

Your Service

Windows Azure Building Blocks

LB

Internet

Web Site(ASPX, ASMX,

WCF)

Web Site(ASPX, ASMX,

WCF)Web Role

(ASPX, WCF)

Worker Service

Worker Role

LB

StorageTables Blobs

Queue

Windows Azure Applications, Storage, and Roles

Cloud Storage (blob, table, queue)

Web RoleLB

n

Worker Role

m

Windows Azure, In One Picture

Desktop

VS Tools

WA SDK

Sto

rage

Clu

ster

MSFT Datacenters

Business Portal

Developer Portal

Com

pute

Clu

ster

……

Service Management Service

REST

REST…

Cloud VM

Runtime API

VM Setup

User Code

Azure

Partial Trust

“Dev Fabric”

Dev Environment

Fabric

Storage

VMs VMs

Windows Azure Compute Service A closer look

…

Load Balanc

er

HTTP/ HTTPS

IIS

WebRole

Instance

WorkerRole

Instance

Compute

Application

Fabric Controlle

r

Fabric Agent

Storage

Fabric Agent

Web RoleInstance

Worker Role

Instance

Windows Azure Fabric

Cloud Simple Application

Web Role Worker Role

Blob

Queue

Windows Azure for Application Developers

Portal HostedService

Storage

SDK

Develop Deploy Run

Roles

Role

Code Configuration

Load Balanced InstancesRole

Instance 1

Instance 2

http://yourapp.cloudapp.net

Web vs Worker Role

Web Role

IIS Host

Your Code

Worker RoleSystem Host

Your Code

Compute

Windows Azure Compute Service

The VMs are provided by a cloud-optimized hypervisor

Not your average VMThey run 64-bit Windows Server 2008Each 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

Storage

Windows Azure Storage

Storage types:Blobs: a simple hierarchy of binary dataTables: entity storage (not relational tables)Queues: high performance quasi-transactional queue structures

Access:Data is exposed via a RESTful interfaceData can be accessed by:

Windows Azure applicationsAny other internet connected applicationon-premises or other cloud applications

…

Fabric

Compute Storage

Application

Windows Azure Storage ServiceA closer look

Blobs

HTTP/ HTTPS

RESTService

s

Tables Queues

Database

Business AnalyticsReporting Data Sync

SQL Services

.NET Services

Windows Azure

Live Services

Applications

Applications

OthersWindows

Mobile

Windows

Vista/XP

Windows

Server

SQL AzureA relational DB in the cloud

SQL Azure Database

Others (Future)

Data Hub

Relational database as a serviceHighly available, automatically maintained

Extension of the SQL Server Data Platform

Database

Relational database, provided as a service

Highly symmetrical development and tooling experience (use TDS protocol and T-SQL)Highly scaled out, on commodity hardwareBuilt on the SQL Server technology foundation

Beyond “hosted database”High availability, DB provisioning, and DB management are provided by the servicePay for what you use

SQL Azure Database

Customer Value Props

Self-provisioning and capacity on demand

Symmetry w/ on-premises database platform

Automatic high-availability and fault-tolerance

Automated DB maintenance (infrastructure)

Simple, flexible pricing – “pay as you grow”

MSDatacente

r

Windows Azure

Compute

Browser

SOAP/RESTHTTP/S

T-SQL (TDS)

App Code(ASP.NET)

SQL Server Report Server

(on-premises)

ADO.NET/REST - EDMHTTP/S

- AD Federation (LiveId /AppFabric AC)

SQL AzureDatabase

T-SQL (TDS)

Relational database service

• SQL Server technology foundation

• Highly symmetrical

• Highly scaled

• Highly secure

Database “as a Service” – beyond hosting

Highly scaled out relational database as a service

Database

Fundamentals

Application Primary FocusDesigning for the Cloud

Scale outServices

AutomatedService

Management

HighAvailability Multi-

tenancy

Resource Optimizatio

n

Transitioning Cloud Driven Process

• Business screening processBusiness Opportunities based on cloud

• Visualization of current statusEnvisioning and Planning

• Business Process suitable for Cloud

Choose Business Case

• Cloud Provider Selection CriteriaCloud implementation

• Public Cloud / Private CloudTransitioning Scenarios

• Systems Resides in the Cloud• Operation OptimizationGoing Production in Cloud

Store Data inthe Cloud

Extend Application Components to the Cloud

Move Entire Applications to the Cloud

Start new applications based

onCloud Services

Create hybrid applications by

Combining Cloud Services

Transitioning to the cloud

57

Migration Paths for Cloud Adoption

Use public cloudsDevelop private clouds

Build a private cloudProcure an outsourced private cloudMigrate data centers to be private clouds (fully virtualized)

Build or procure community cloudsOrganization wide SaaSPaaS and IaaSDisaster recovery for private clouds

Use hybrid-cloud technologyWorkload portability between clouds

Infrastructure Transition

Cloud Reference Architecture

Cloud Computing for Financial Institutions. Vivek Kejriwal

Transitioning Cloud Driven Architectures




Application Architecture for the Cloud

65

Application Basic Building Blocks

Cloud Storage (blob, table, queue)

Presentation RoleLB

n

Service Role

m

N-Tier Application Transition

Presentation

Business

Data Access

Data

Move to the cloud On Premises

Data


Presentation

Business

Data Access

Data

Move to the cloud On Premises

Data


Presentation

Business

Data Access

Data On Premises Data

Move to the cloud

Data

Move to the cloud

Sync

Simple Cloud Application

Presentation Services

Blob

Queue

Sample Cloud Application Arquitecture

Cloud Relational Storage

Common DB

Tables

Castellan Venue

DB


ShardingPartition

Castellan Venue

DB


ShardingPartition

Castellan Venue

DB


ShardingPartition

...

Cloud Based Roles

http:// sophia.* Dynamic Worker

(blobs and sharding

management)

Partitioner Worker

Cloud StorageQueues for communication between clients and roles

-- - --- - -

-- - --- - -

Tables to record server & partition information

Blobs to store web and worker role resources

Client Applications

Application on Premises

Multimedia Application• Common Educational Experiences

Communication Services

Communication Services

On PremisesData Storage

Configuration Statistics

Distributed Cache Worker

Access Control

Service Bus

Service Bus

Registry

Endpoints

Organization Y

Organization X

Application Application

Cloud Communication Pattern

2) Discover endpoints

1) Register endpoints3) Access

application

4) Upgrade to direct

connection

Transitioning to the CloudDesign Requirements

LooseCoupling

Component

Re-use

Scalability

Principle Details

Loose coupling • Messaging communication

• Component Autonomy• Component

Interoperability

Scalability • Design for partitioning and elasticity

• Stateless component design

• Consider long running processes

Component Re-use • Composition • Integration

Consider cluster design

• Split the application in multiple clusters by functionality

Basic cloud development principles

Principle Details

Failure handling • Can some parts of the application continue to execute without the other parts

Data persistence • Storage and replication

SOA is reinforced • More loosely coupled the components of the system, the better it scales

• Asynchronous systems

Parallel design • Processing and storage queries

Security • Build security on every layer

Basic cloud development principles

Cloud Transitioning ArchitectureIdentify PatternsConsider:

Basic componentsMessagingInteroperable Message Format and ProtocolsCommunication or Integration BusResource Deployment

StorageInfrastructure (Processing, Creation, …)

Scaling Out using a Queue Based Pattern

Queue

Msg 1

Msg 2

Msg 3

Msg 4

Process Role

Worker Role

Submit Message

Application

Process Message

Msg 2Msg 1

Process Role

Msg 2

AlarmProcessor

IncidentTable

Pos RptTable

TimerTable

EventTable

NewMsg

Parse & Route

Queue A

Implement Queues for Reliability and Continuity

Processor A

Table

Table

Table

Queue B

Processor B

Building Blocks Key takeaways

Use the cloud provider service configuration to scaleDesign how to communicate your application componentsBuilt your application to scale up/down roles in production

Windows Azure Storage Service

Windows Azure Storage ServiceA closer look

Fabric

…

Compute

Application

Storage

Blobs Tables Queues

HTTP/ HTTPS

Fundamental Storage Abstractions

Tables – Provide structured storage. A Table is a set of entities, which contain a set of properties

Queues – Provide reliable storage and delivery of messages for an application

Blobs – Provide a simple interface for storing named files along with metadata for the file

Drives – Provides durable NTFS volumes for Windows Azure applications to use (new)

81

Windows Azure StorageTables: Strengths

Massive scalabilityBy effectively allowing scale-out data

Perspective:Applied to the right problem, Windows Azure Tables are a beautiful thing

But they’re not the optimal solution for a majority of data storage scenarios

Amazon, Google, and others provide similar cloud storage mechanismsAll of these can be seen as part of the NOSQL movement

Windows Azure StoragePoints of interest

Storage types:Blobs: store binary dataTables: provide scale-out, entity-based storage

Not relational tables

Queues: allow message-based communication

Access:Data is grouped into storage accountsData can be accessed by:

Windows Azure apps Other on-premises or cloud apps

Windows Azure StorageA closer look at blobs

Options for accessing blobs:From Windows Azure applications or other applications: via RESTful requestsFrom Windows Azure applications: via Windows Azure Drives

Drives make a blob look like an NTFS file system to the application

Blobs can also be used with the Windows Azure Content Delivery Network (CDN)

For faster access from distributed locations to frequently accessed content

Windows Azure StorageA closer look at tables

Name Type Value

Table . . .Table Table

Storage Accounts

Entity . . .Entity Entity

Property PropertyProperty

Windows Azure TablesProvides Structured Storage

Massively Scalable TablesBillions of entities (rows) and TBs of dataCan use thousands of servers as traffic grows

Highly Available & DurableData is replicated several times

Familiar and Easy to use APIADO.NET Data Services – .NET 3.5 SP1

.NET classes and LINQREST – with any platform or language86

Windows Azure StorageTables: Challenges

No SQLA new approach for developers to learnNo aggregates

Access via RESTYou can’t use ordinary ADO.NET

An unfamiliar structure for dataYou can’t easily move relational data in or outSupporting services are scarce, e.g., reporting

No schemaAnd no views

Queue

Windows Azure StorageA closer look at queues

Web Role

ASP.NET, WCF, etc.

Worker Role

main(){ … }

1) Receive work

2) Put message in

queue

3) Get message

from queue

4) Do work

5) Delete message

from queue

Windows Azure Tables

Windows Azure TablesProvides Structured Storage

Massively Scalable TablesBillions of entities (rows) and TBs of dataCan use thousands of servers as traffic grows

Highly Available & DurableData is replicated several times

Familiar and Easy to use APIADO.NET Data Services – .NET 3.5 SP1

.NET classes and LINQREST – with any platform or language90

Table Storage ConceptsEntitiesTablesAccounts

moviesonline

Users

Movies

Email =…Name = …

Email =…Name = …

Genre =…Title = …

Genre =…Title = …

91

Table Data ModelTable

A storage account can create many tablesTable name is scoped by accountSet of entities (i.e. rows)

EntitySet of properties (columns)Required properties

PartitionKey, RowKey and Timestamp

92

Required Entity PropertiesPartitionKey & RowKey

Uniquely identifies an entityDefines the sort orderUse them to scale your application

Timestamp Read onlyOptimistic Concurrency

93

PartitionKey And PartitionsPartitionKey

Used to group entities in the table into partitions

A table partition All entities with same partition key valueUnit of scaleControl entity localityRow key provides uniqueness within a partition

94

PartitionKey(Category)

RowKey(Title)

Timestamp ReleaseDate

Action Fast & Furious … 2009

Action The Bourne Ultimatum

… 2007

… … … …

Animation

Open Season 2 … 2009

Animation

The Ant Bully … 2006PartitionKey(Category)

RowKey(Title)

Timestamp

ReleaseDate

Comedy Office Space … 1999

… … … …

SciFi X-Men Origins: Wolverine

… 2009

… … … …

War Defiance … 2008


RowKey(Title)

Timestamp ReleaseDate

Action Fast & Furious … 2009


… 2007

… … … …

Animation

Open Season 2 … 2009

Animation

The Ant Bully … 2006

… … … …

Comedy Office Space … 1999

… … … …

SciFi X-Men Origins: Wolverine

… 2009

… … … …

War Defiance … 2008

Partitions and Partition Ranges

Server BTable = Movies

[Comedy- Western)

Server ATable = Movies[Action - Comedy)

95

Server ATable = Movies

Table OperationsTable

CreateQueryDelete

EntitiesInsertUpdate

Merge – Partial UpdateReplace – Update entire entity

DeleteQueryEntity Group Transaction (new)

Table SchemaDefine the schema as a .NET class

97

[DataServiceKey("PartitionKey", "RowKey")] public class Movie { /// <summary> /// Category is the partition key /// </summary> public string PartitionKey { get; set; }

/// <summary> /// Title is the row key /// </summary> public string RowKey { get; set; }

public DateTime Timestamp { get; set; }

public int ReleaseYear { get; set; } public string Language { get; set; } public string Cast { get; set; } }

Table SDK Sample Code

98

StorageCredentialsAccountAndKey credentials = new StorageCredentialsAccountAndKey( “myaccount", “myKey");string baseUri = "http://myaccount.table.core.windows.net";

CloudTableClient tableClient = new CloudTableClient(baseUri, credentials);

tableClient.CreateTable(“Movies");

TableServiceContext context = tableClient.GetDataServiceContext();CloudTableQuery<Movie> q = (from movie in context.CreateQuery<Movie>(“Movies")

where movie.PartitionKey == “Action" && movie.RowKey == "The Bourne Ultimatum"

select movie).AsTableServiceQuery<Movie>();Movie movieToUpdate = q.FirstOrDefault();

// Update moviecontext.UpdateObject(movieToUpdate);context.SaveChangesWithRetries();

// Add moviecontext.AddObject(new Movie(“Action" , movieToAdd));context.SaveChangesWithRetries();

Key Selection: Things to Consider> Scalability

> Distribute load as much as possible> Hot partitions can be load balanced> PartitionKey is critical for scalability

> Query Efficiency & Speed> Avoid frequent large scans> Parallelize queries

> Entity group transactions (new)> Transactions across a single partition> Transaction semantics & Reduce round trips

99

Key Selection: Case Study 1> Table for listing all movies

> Home page lists movies based on chosen category

100

Movie Listing – Solution 1> Why do I need multiple PartitionKeys?

> Account name as Partition Key > Movie title as RowKey since movie names

need to be sorted> Category as a separate property

> Does this scale?

101

PartitionKey(Account name)

RowKey(Title)

Category …

moviesonline 12 Rounds Action …

moviesonline A Bug’s Life

Animation …

100,000,000 more rows

… … …

moviesonline Office Space

Comedy …

moviesonline Platoon War …

50,000,000 more rows

… … …

moviesonline WALL-E Animation …

Server A

Movie Listing – Solution 1

> Single partition - Entire table served by one server

> All requests served by that single server> Does not scale

PartitionKey(Account name)

RowKey(Title)

Category …

moviesonline 12 Rounds Action …

moviesonline A Bug’s Life

Animation …

100,000,000 more rows

… … …

moviesonline Office Space

Comedy …

moviesonline Platoon War …

50,000,000 more rows

… … …

moviesonline WALL-E Animation …

ClientClient Request

Request

Request

Request

102

Movie Listing – Solution 2


RowKey (Title)

Action Fast & Furious

… 10000 more Action movies


… 100000 more Action & Animation movies

Animation Open Season 2

… 100000 more Animation movies

Animation The Ant Bully

Comedy Office Space

… 1000000 more Comedy & SciFi movies

SciFi Star Trek

… 100000 more SciFi & War movies

… 100000 more War movies

War Defiance

> All movies partitioned by category> Allows system to load balance hot partitions> Load distributed> Better than single partition

ClientClient Request

Server A

Request

Request

Request

103

Server B

Request

Request

Request

Request


RowKey (Title)








Comedy Office Space


SciFi Star Trek



War Defiance

Key Selection: Case Study 2

> Log every transaction into a table for diagnostics> Scale Write Intensive Scenario> Logs can be retrieved for a given time

range

104

Logging - Solution 1> Timestamp as Partition Key

> Looks like an obvious choice> It is not a single partition as time moves

forward > Append only> Requests to single partition range> Load balancing does not help> Server may throttle

PartitionKey(Timestamp)

Properties

2009-11-15 02:00:01

…

2009-11-15 02:00:11

…

100000 more rows …

2009-11-17 05:40:01

…

2009-11-17 05:40:01

…

80000 more rows …

2009-11-17 12:30:00

…

2009-11-17 12:30:01

…

ApplicationsClientReques

t

Server A

Request

2009-11-17 12:30:01

…

Request

2009-11-17 12:30:02

…

Request

2009-11-17 12:30:03

…

Server B

105

Server A

Server B

PartitionKey(ID_Timestamp)

Properties

01_2009-10-12 05:10:00

…

… …


09_2009-11-15 12:31:00

…

… …


10_2009-10-05 05:10:10

…


… …


19_2009-11-17 12:20:02

…

ApplicationsClientReques

tReques

tReques

tReques

t

Logging Solution 2 - Distribute "Append Only”

> Prefix timestamp such that load is distributed> Id of the node logging> Hash into N buckets

> Write load is now distributed > Better throughput> To query logs in time range

> Parallelize it across prefix values

15_2009-11-17 12:30:01

…

09_2009-11-17 12:30:22

…

19_2009-11-17 12:30:10

…

01_2009-11-17 12:30:01

…

106

Key Selection: Query Efficiency & Speed> Select keys that allow fast retrieval> Reduce scan range> Reduce scan frequency

107

Single Entity Query

Server A


RowKey (Title)








Comedy Office Space


SciFi Star Trek



War Defiance

Client

Server B

> Where PartitionKey=‘SciFi’ and RowKey = ‘Star Trek’

> Efficient processing> No continuation tokens

108

Request

Result

Client

Server A

Server B

Table Scan Query

Request


RowKey(Title)

Rating

Action Fast & Furious 5

… 999 more movies rated > 4

…

… Action and Anim. movies here with rating < 4

…

Animation A Bug’s life 2

… 100 more movies < 4 here

…

Animation The Ant Bully 3

Comedy Are we there yet? 2

… More movies here …

Comedy Office Space 5

… 800000 more movies here

…

Drama A Beautiful Mind 5

… 1200000 more movies here

…

War Defiance 4

Cont.

> Select * from Movies where Rating > 4> Returns Continuation token

> 1000 movies in result set> Partition range boundary

> Serial Processing: Wait for continuation token before proceeding

Request Cont.

Cont.

Request Cont.

Cont.

109

Returns 1000 movies

Partition range boundary hit

Return continuation

Client

Server A

Server B

Make Scans Faster

Request


RowKey(Title)

Rating

Action Fast & Furious 5


Comedy Office Space 5


Documentary

Planet Earth 4

… More movies here

Drama Seven Pounds 4

Horror Saw 5 3


Music 8 Mile 2


SciFi Star Trek 5


Cont.

> Split “Select * from Movies where Rating > 4” into> Where PartitionKey >= “A” and PartitionKey < “D” and Rating > 4> Where PartitionKey >= “D” and PartitionKey < “I” and Rating > 4> Etc.

> Execute in parallel> Each query handles continuation

Cont.

110

Request

Request

Cont.

Query Speed1. Fast

> Single PartitionKey and RowKey with equality

2. Medium> Single partition but a small range for RowKey> Entire partition or table that is small

3. Slow> Large single scan> Large table scan> “OR” predicates on keys => no query

optimization => results in scan> Expect continuation token for all except in 1

111

Make Queries Faster> Large Scans

> Split the range and parallelize queries> Create and maintain own views that help

queries

> “Or” Predicates> Execute individual query in parallel

instead of using “OR”

> User Interactive> Cache the result to reduce scan

frequency

112

Expect Continuation Tokens – Seriously!> Maximum of 1000 rows in a response> At the end of partition range boundary> Maximum of 5 seconds to execute the

query

113

Entity Group Transactions (EGT) (new)> Atomically perform multiple

insert/update/deleteover entities in same partition in a single transaction

> Maximum of 100 commands in a single transaction and payload < 4 MB

> ADO.Net Data Service> Use SaveChangesOptions.Batch

114

Key Selection: Entity Group Transaction> Case Study

> Maintain user account information> Account ID, User Name, Address, Number of rentals

> Maintain information of checked out rentals> Account ID, Movie Title, Check out date, Due date

> Solution 1 – Maintain two tables – Users & Rentals > Handle Cross table consistency

> Insert into Rentals table succeeds> Update to Users table fails> Queue to maintain consistency

115

Solution 2> Store Account Information and Rental details in

same table> Maintain same PartitionKey to enforce transactions

> Account ID as PartitionKey> Update total count and Insert new rentals using Entity

Group Transaction> Prefix RowKey with “Kind” code: A = Account, R = Rental

> Row key for account info: [Kind Code]_[AccountId]> Row Key for rental info: [Kind Code]_[Title]

> Rental Properties not set for Account row and vice versaPartitionKey(AccountID)

RowKey(Kind_*)

Kind TotalRentals

Name Address CheckOutOn Title DueOn

… … … … … … … … …

Sally A_Sally Account

8 Sally Field

Ann Arbor, MI

Sally R_Jaws Rental 2009/11/16 Jaws 2009/11/20

Sally R_Taxi Rental 2009/11/16 Taxi 2009/11/20

… … … … … … … … …116

Best Practices> Select PartitionKey and RowKey that help scale

> Efficient for frequently used queries> Supports batch transactions> Distributes load

> Distribute “Append only” patterns using prefix to PartitionKey

> Always Handle continuation tokens

> Client can maintain their own cache/views instead of frequent scans> Future Feature - Secondary Index

> Execute parallel queries instead of “OR” predicates

> Implement back-off strategy for retries

117

Windows Azure Queues

Queue are performance efficient, highly available and provide reliable message delivery

Simple, asynchronous work dispatchProgramming semantics ensure that a message can be processed at least once

Access is provided via REST

118

Queue Storage ConceptsMessage

sQueuesAccounts

sally

thumbnailjobs

traverselinks

128 x 128 http://...

256 x 256 http://...

http://...

http://...

119

Account, Queues and MessagesAn account can create many queues

Queue Name is scoped by the account

A Queue contains messagesNo limit on number of messages stored in a queueSet a limit for message expiration

MessagesMessage size <= 8 KBTo store larger data, store data in blob/entity storage, and the blob/entity name in the messageMessage now has dequeue count 120

Queue OperationsQueue

Create QueueDelete QueueList QueuesGet/Set Queue Metadata

MessagesAdd Message (i.e. Enqueue Message)Get Message(s) (i.e. Dequeue Message)Peek Message(s)Delete Message

121

Queue Programming Api

122

CloudQueueClient queueClient = new CloudQueueClient(baseUri, credentials);CloudQueue queue = queueClient.GetQueueReference("test1");

queue.CreateIfNotExist();

//MessageCount is populated via FetchAttributesqueue.FetchAttributes();

CloudQueueMessage message = new CloudQueueMessage("Some content");queue.AddMessage(message);

message = queue.GetMessage(TimeSpan.FromMinutes(10) /*visibility timeout*/);

//Process the message here …

queue.DeleteMessage(message);

21

11

C1

C2

Removing Poison Messages

11

21

340

Producers Consumers

P2

P1

30

2. GetMessage(Q, 30 s) msg 2

1. GetMessage(Q, 30 s) msg 1

11

21

123

10

20

C1

C2


340

Producers Consumers

P2

P1

11

21

2. GetMessage(Q, 30 s) msg 23. C2 consumed msg 24. DeleteMessage(Q, msg 2)7. GetMessage(Q, 30 s) msg 1

1. GetMessage(Q, 30 s) msg 15. C1 crashed

11

21

6. msg1 visible 30 s after Dequeue30

124

12

11

12

C1

C2


340

Producers Consumers

P2

P1

12

2. Dequeue(Q, 30 sec) msg 23. C2 consumed msg 24. Delete(Q, msg 2)7. Dequeue(Q, 30 sec) msg 18. C2 crashed

1. Dequeue(Q, 30 sec) msg 15. C1 crashed10. C1 restarted11. Dequeue(Q, 30 sec) msg 112. DequeueCount > 213. Delete (Q, msg1)

12

6. msg1 visible 30s after Dequeue9. msg1 visible 30s after Dequeue

30

125

13

12

13

Best Practices & Summary> Make message processing idempotent

> No need to deal with failures

> Do not rely on order> Invisible messages result in out of order

> Use Dequeue count to remove poison messages> Enforce threshold on message’s dequeue count

> Use message count to dynamically increase/reduce workers

> Use blob to store message data with reference in message> Messages > 8KB> Batch messages> Garbage collect orphaned blobs

126

Storage Summary

127

> Table> Scalable & Reliable Structured Storage System> Partitioning is critical to scalability> Entity Group Transactions (new)

> Queue> Scalable & Reliable Messaging System > Dequeue count returned with message (new)

> Use back-off strategy on retries

> Official Storage Client Library (new)

What is “SQL Azure”?

Review – Conceptual modelSubscription

Used to map service usage to the billing instrumentUsers may have many subscriptions

Logical ServerAkin to SQL Server InstanceUnit of Geo-Location & Billing1:1 Subscription & server

User DatabaseRestricted T-SQL surface areaAdditional catalog views provided e.g. sys.billing, sys.firewall_rules, etc

SQL Services

.NET Services

Windows Azure

Live Services

Applications

Applications

OthersWindows

Mobile

Windows

Vista/XP

Windows

Server

SQL AzureA relational DB in the cloud

SQL Azure Database

Others (Future)

Data Hub

Relational database as a serviceHighly available, automatically maintained

Extension of the SQL Server Data Platform

SQL Azure Database

- Based on SQL Server 2008 R2 engine

Use same tools and data access frameworksSix global datacentersHigh Availability & Redundancy

Reads are completed at the primaryWrites are replicated to a quorum of secondaries

Replica 1

Replica 2

Replica 3

DB

Single LogicalDatabase

Multiple PhysicalReplicas

Single Primar

y

MultipleSecondaries

Extending SQL Server Data Platform to the Cloud

Database

Data Sync

Reference

Data

Symmetric Programming Model

Data Hub Aggregation

• Initial services – core RDBMS capabilities with SQL Azure Database, Data Sync

• Future Offerings• Additional data platform capabilities: Reporting, BI• New services: Reference Data

The New SQL Data Services

Familiar SQL Server relational modelUses existing APIs & toolsBuilt for the Cloud with availability and scaleAccessible to all from PHP, Ruby, and Java

Clear Feedback: “I want a database in the Cloud”

Focus on combining the best features of SQL Server running at scale with low friction

The Evolution of SDS

Evolv

es

SDS Next

TDS + TSQL Model

Web App

SQL Client*

Windows Azure

BrowserApplicatio

n

Application

REST Client

REST (Astoria)

ADO.Net + EF

Application

SQL Client*

Cloud

HT

TP

HT

TP

+R

ES

T

TD

S

* Client access enabled using TDS for ODBC, ADO.Net, OLEDB, PHP-SQL, Ruby, …

Data

Cen

ter

ODBC, OLEDB, ADO.Net PHP, Ruby, …

SDS Current

REST/SOAP + ACE Model

Web App

REST Client

Windows Azure

BrowserApplicatio

n

Application

REST Client

HT

TP

HT

TP

+R

ES

T

Data

Cen

ter

Cloud

SQL Azure

Windows Azure

Applications

Applications

OthersWindows

SQL Azure Database

Others (Future)

SQL Azure


SQL Azure Database

Others (Future)

SQL Azure DatabaseAn Illustration

Database

Database

Database

SQL Azure Database

TDS

Application

Database

Database

Database

Application

ApplicationDatabase

SQL Azure Database

SQL Azure DatabaseUsing one or multiple databases

Max database size is 50

GB

SQL Azure Network TopologyApplication

InternetAzure Cloud

LB

TDS (tcp)

TDS (tcp)

TDS (tcp)

Applications use standard SQL client libraries: ODBC, ADO.Net, PHP, …

Load balancer forwards ‘sticky’ sessions to TDS protocol tier

Security Boundary

SQL SQL SQL SQL SQLSQL

Gateway Gateway Gateway Gateway Gateway Gateway

Gateway: TDS protocol gateway, enforces AUTHN/AUTHZ policy; proxy to CloudDB

Scalability and Availability: Fabric, Failover, Replication, and Load balancing

TDS Gateway

TDS ListenerCapability negotiationTDS Packet inspectionSecurity

Logical->Physical mapping via metadata catalogEnabler for multi-tenet capabilitiesIsolation layer

TDS Gateway LayeringGateway Process

TDS EndpointAdminSvc Endpoint

Protocol Parser

Provisioning Endpoint

Business Logic Services

Connection Mgmt

SQL SQL SQL SQL SQLSQL

Scalability and Availability: Fabric, Failover, Replication, and Load balancing

Provisioning

SubscriptionCoordinated across all Azure servicesExecuted in parallel w/retries

ServerMay occur between data centersPoint where Geo-location is established

DatabaseAlways occurs within a single data centerCross node operations executed during this process e.g. add new db to sys.databases on the master

Server ProvisioningDriven by administrator PortalProvision request is sent to Gateway

Metadata catalog entry createdDNS record (CNAME) created within LiveDNS serviceMaster DB created

On completion metadata catalog updated

Datacenter (Sub-Region)

SQL Azure Server Provisioning

Customer Browser

Portal LB

1

Front-end Node

Live DNS ClusterLive DNS

Svc

2 4

Gateway

Backend Node

Mgmt. Service

s

SQL Server

Fabric

Backend Node Backend Node

Mgmt. Service

s

SQL Server

Fabric

Mgmt. Service

s

SQL Server

Fabric

Front-end Node

Gateway

Front-end Node

Admin Portal

Front-end Node

Admin Portal

3

Gateway LB5

6

7

Database Provisioning

Gateway performs stateful TDS packet inspection

Picks out subset of messagesParses out args for create database

Makes entry into Gateway metadata catalogUnused replica set located and reservedReplica set (UserDB) is prepped for use Metadata catalog is updated

SQL Azure Database provisioningTDS Gateway

Front-end NodeProtocol Parser

Gateway Logic

Master Cluster

Master Node

Master Node Components

1

2

3

5 67

Scalability and Availability: Fabric, Failover, Replication, and Load balancingScalability and Availability: Fabric, Failover, Replication, and Load balancing

4

8

TDS Session

Backend Node 1

SQL InstanceSQL DB

Backend Node 2

SQL InstanceSQL DB

Backend Node 3

SQL InstanceSQL DB

SQL Azure Login Process

Login request arrives at the GatewayGateway locates MasterDb & UserDb replica setsCredentials are validated against MasterDbTDS session is opened to UserDB and requests are forwarded

SQL Azure Login ProcessTDS GatewayFront-end Node

Protocol Parser

Gateway Logic

Global Partition Map

Master Node

Master Node Components

1

2

4 5

6

Scalability and Availability: Fabric, Failover, Replication, and Load balancingScalability and Availability: Fabric, Failover, Replication, and Load balancing

3

7

TDS Session

Backend Node 1

SQL InstanceSQL DB

Backend Node 2

SQL InstanceSQL DB

Backend Node 3

SQL InstanceSQL DB

8

Service Resilience

ProvisioningState machines used to coordinate activities across node (and datacenter) boundariesFailed provisioning attempts cleaned automatically after 10 minutes

LoginFailovers during the login will be transparent (<30 seconds)Metadata catalog refresh occurs automatically

Active SessionSurface as connection drops (due to state)

Monitoring Service HealthMetrics

Cluster wide performance counters gather key metrics on the serviceUsed to alert Operations to issues before they become a problem

Early warning systemCode issuesCapacity warnings

HealthExercises the service routinely looking for problemsWhen issues are encountered runs deep diagnostics

Network connectivity at the node levelValidate all dependent services (Live DNS, Live ID, etc)

Monitoring from other MSFT DC’sValidates accessibility from multiple geographic locationsAlerts fired automatically when test jobs fail

Security/Attack ConsiderationsService

Secure channel required (SSL)Denial Of Service trend trackingPacket Inspection

Server IP allow list (Firewall) Idle connection cullingGenerated server names

DatabaseDisallow the most commonly attacked user id’s (SA, Admin, root, guest, etc) Standard SQL Authn/Authz mode

What’s New In SQL Azure CTPEnhancements

SQL Azure Database

SQL Azure Reporting CTPReporting and BI on SQL Azure databasesBased on SQL Server Reporting Services

SQL Azure Data Sync CTP2Synchronize SQL Azure databasesSynchronize SQL Azure with on-premises SQL Server

Windows Azure Platform – Developer Portal

Enhanced user experience and reportingDatabase manager for SQL Azure


SQL Azure Database




Enhanced user experience and reportingDatabase manager for SQL Azure


SQL Azure Database




Enhanced user experience and reportingDatabase management

SQL Azure Reporting CTP

“Introduction to SQL Azure Reporting”

Web Role

Reporting

Features:

Based on SQL Server Reporting ServicesInteractive and tabular reportingData visualizations: charts, graphs, mapping, gauges

SQL Azure Database data sourcesUtilizes BI Developer Studio (free) to author reportsExport to Excel, PDF, CSV formats

Core scenarios:

Operational reporting on SQL Azure dataEmbed reports into Windows Azure or on-premises applications

http://www.adobe.com/go/getreader/

What’s New In SQL AzureEnhancements

SQL Azure Database




Enhanced user experience and reportingDatabase management

SQL Azure Data Sync CTP2

Elastic ScaleService scales as resources requirements grow

No-Code Sync ConfigurationEasily define data to be synchronized

Schedule SyncChoose how often data is synchronized

Conflict HandlingHandle issues where same data is changed in multiple locations

Logging and MonitoringAdministration capabilities for tracking data and monitoring potential issues

“Introduction to SQL Azure Data Sync”

CTP 1

CTP 2

On Premises Windows Azure

Data Sync Service

SQL Azur

e

TDS

SQL Serve

r

Local Agent

SQL Server Sync

Provider

SQL Server Proxy

Provider

Sync Orchestrator

SQL ServerProvider

Sync Orchestrator

HTTPS

SQL Azure Data Sync – A Closer LookCTP 2

SQL Azure Data Sync – Example Use Cases

Move workloads in stages preserving existing infrastructureMove part of the application and sync its data

Meet compliance and regulationsControl data synchronized off-premises

Enable scale-out read or read/writeMultiple synchronized databases for scalability

Preserve data – geo replication of data

Enable new scenariosSpanning enterprise, cloud and remote offices/retail stores

SQL Azure Data Sync – Roadmap

On-Premises (Headquarters)

Syn

c

Sync

Remote OfficesData Sync Service For SQL Azure

Retail Stores

Sync Sync

Sync

SQL Azure

Database

Sync Sync Now

CTP2

– E

nd

of

CY1

0

New SQL Azure Usage Scenarios

Web Role

Reporting

Data SyncFully featured

Windows Azure PlatformApplication


New SQL Azure Usage Scenarios

Web Role

Reporting

Data Sync

Fully featured Windows Azure Platform

Application

Data Sync

Synchronize multiple on-premises SQL Server databases in different

locations

Datacenter

Corporate Office

Branch Office or Retail

Geo-replication

of data across

Windows Azure

Platform datacente

rs

Globally available,cloud & on-premisesdata synchronization


RoadmapNow

Visit www.microsoft.com/sqlazure to register for upcoming CTP access

2010New Windows Azure Developer Portal; Database ManagerLimited CTPs of Data Sync CTP2, Reporting CTPWatch for PDC Announcements; visit the SQL Azure blog and website

2011Backup & RestoreSQL Azure ReportingSQL Azure Data Sync

http://www.microsoft.com/sqlazure

Windows Azure AppFabricInfrastructure in the cloud

SQL Azure

Windows Azure

Applications

Applications

OthersWindows

Service Bus

?Access Control


Access Control

Service Bus

Service Bus

Registry

4) Invoke operation on Service Bus

endpoint

3) Discover Service Bus endpoint

Client (On-

premises or cloud)

WCF Service(On-premises)

2) Expose Service Bus endpoint

Illustrating Service Bus

5) Invoke operation on service endpoint

1) Register service endpoint

Windows Azure Platform PricingA summary (in US dollars)Compute: $0.12 to $0.96/hour for each

role instance depending on instance size Storage:

Windows Azure blobs and tables: Data: $0.15/GB per monthAccess: $0.01/10,000 operations

SQL Azure relational:$9.99/month per GB

Bandwidth: Americas/Europe: $0.10/GB in, $0.15/GB outAsia/Pacific: $0.30/GB in, $0.45/GB out

Service Level AgreementsKey points: Windows Azure compute

When you deploy two or more role instances in different fault and upgrade domains, your Internet-facing roles will have external connectivity at least 99.95% of the timeMicrosoft will detect within two minutes when a role instance’s process is not running and initiate corrective action

Service Level AgreementsKey points: Windows Azure storage

At least 99.9% of the time, Microsoft will successfully process correctly formatted requests to add, update, read and delete dataAt least 99.9% of the time, your storage accounts will have connectivity to Microsoft’s Internet gateway

Service Level Agreements Key points: SQL Azure database

SQL Azure will maintain a Monthly Availability of 99.9% during a calendar month. Monthly Availability percentage for a specific customer database is the ratio of the time the database was available to customer to the total time in a month. An interval is marked as unavailable if the customer’s attempts to connect to a database are rejected by the SQL Azure gateway.

Using the Windows Azure Platform: Scenarios

Applying Azure (1) Some characteristics of Azure-appropriate applications

Apps that need massive scaleExample: A Web 2.0 application

Apps that need high reliabilityExample: A Software as a Service (SaaS) application

Apps with variable loadExample: An on-line ticketing application

Apps with a short or unpredictable lifetime

Example: An app created for a marketing campaign

Applying Azure (2) Some characteristics of Azure-appropriate applicationsApps that do parallel processing

Example: A financial modeling applicationApps that must fail fast or scale fast

Example: Start-upsApps that don’t fit well in an organization’s data center

Example: A business unit that wishes to avoid its IT department

Apps that can benefit from external storage

Example: An application that archives data

Blobs

Building on AzureA non-Azure app using Windows Azure storage

SQL Azure

On-Premises or

Hosted Application

Web RoleInstance

Tables

Building on Azure A massively scalable web app exposed on the public Internet

Users

BlobsQueuesTables

Users

Worker Role

Instance

Web RoleInstance

Building on Azure A massively scalable web app with background processing

Web RoleInstance

SQL Azure

Users

Building on Azure A web app with relational storage

Web RoleInstance

SQL Azure

Users

Building on Azure A web app using cloud and on-premises data

Service Bus

On-premises Database

User

Blobs

Web RoleInstance

Queues

Worker Role

Instance

Building on Azure A parallel processing application


AppFabric

Developer ExperienceUse existing skills and tools.

Compute Storage Management

Relational data Management Connectivity Access control

platform

CDN

Conclusions

Cloud platforms are hereMicrosoft is placing a big bet with the Windows Azure platform

A new world is unfoldingPrepare to be part of it

Related ContentComunidad Windowshttp://comunidadwindows.orghttp://www.slideshare.net/ecastromhttp://ecastrom.blogspot.com

Introducing the Windows Azure Platform

http://go.microsoft.com/fwlink/?LinkId=158011

Introducing Windows Azurehttp://go.microsoft.com/?linkid=9682907

Presentation SourcesResources

SQL Azure website: http://www.microsoft.com/sqlazure

Blog: http://blogs.msdn.com/sqlazure

Dev Center: http://msdn.microsoft.com/windowsazure/sqlazure

Frameworks:http://msdn.microsoft.com/data

http://www.microsoft.com/sqlazure

http://blogs.msdn.com/sqlazure

http://msdn.microsoft.com/windowsazure/sqlazure

http://msdn.microsoft.com/windowsazure/sqlazure

http://msdn.microsoft.com/data

An Introduction to the Windows Azure PlatformDr. Eduardo Castro MartínezMicrosoft [email protected]

http://comunidadwindows.orghttp://ecastrom.blogspot.com

© 2010 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.

Technology

Cloud application architecture with sql azure and windows azure