Patterns for Cloud Computing

Patterns for Cloud Computing

Simon Guest
Senior Director, Technical Strategy
Microsoft Corporation


This is Jim


Jim has many questions about cloud computing


The more he reads, the more confused he gets


“What is cloud computing?”


“What are vendors doing in this space?”


“What applications make sense in the cloud?”

Session Objectives

Build on Stefan’s introduction of
of cloud computing

Provide you with 5 patterns for
cloud-based applications

Show implementations of
these patterns

Defining Cloud Computing

Application runs
on-premises

• Bring my own
machines,
Buy my own
connectivity,
software, etc.and
hardware,
• manage my own
Complete control
data center
and responsibility
• Upfront capital
costs for the
infrastructure


Application runs Application runs at a
on-premises hoster

• Bring my own • Rent machines,
machines, connectivity,
Buy my own Pay someone to
software
connectivity,
software, etc.and
hardware, host my application
• Less control, but
• manage my own
Complete control using hardware that
fewer
data center
and responsibility I specify
responsibilities
• Upfront capital • Lower capital costs,
costs for the but pay for fixed
infrastructure capacity, even if idle


Application runs Application runs at a Application runs using
on-premises hoster cloud platform

• Bring my own • Rent machines, • Shared,
machines, connectivity, multi-tenant
Pay someone for a
Buy my own Pay someone to
software environment
connectivity, pool of computing
software, etc.and
hardware, host my application
• Less control, but • Offers pool of
resources that can
• manage my own
Complete control using hardware that
fewer computing
be applied to a set
data center
and responsibility I specify
responsibilities resources,
of applications
abstracted from
• Upfront capital • Lower capital costs,
costs for the but pay for fixed infrastructure
infrastructure capacity, even if idle • Pay as you go


Cloud “Variants”


Public Cloud


Pool of computing resources offered by
a vendor, typically using a
“pay as you go” model


Private Cloud


Pool of computing resources that lives
within a self managed datacenter


Pool of computing resources that lives
within a datacenter with no sharing


Compute: Virtualized compute based on Windows Server
Storage: Durable, scalable, & available storage
Management: Automated, management of the service

Database: Relational processing for structured/
unstructured data

Service Bus: General purpose application bus
Access Control: Rules-driven, claims-based access control


Different Models

Infrastructure as a Service (IaaS)
vs.
Platform as a Service (PaaS)


Your Application Deployment

Frameworks

Deployment Web Server

OS Services Provided by
Windows Azure
Operating System

Provided Virtualized Instance
By
Amazon
Hardware
EC2


Deployment Your Application Deployment

Frameworks

Web Server

Provided by OS Services Provided by
Google
Windows Azure
AppEngine
Operating System

Virtualized Instance

Hardware


#1 - Using the Cloud for Scale


“Isn’t the cloud good for applications that need to
scale dynamically?”


“For example, applications that have spikes or
peak loads”


“How does this work?”


Let’s do some white boarding for Jim…

…and when I say white boarding, I really mean it


“Wow! What a great site!”

Request
Browser Web Tier B/L Tier Database
Response


Browser

Browser

“Server Busy”
Browser

Browser


Browser

Browser


Browser “Timeout”

Browser

How would Jim do this today on premises?


Browser

Browser Web Tier
N
Browser L Web Tier B/L Tier Database
B
Browser Web Tier

Browser


Browser

Browser B/L Tier
N
Browser Web Tier L B/L Tier Database
B
Browser B/L Tier

Browser


Browser

Browser Web Tier B/L Tier
N N
Browser L Web Tier L B/L Tier Database
B B

Browser


Browser p1 p2 p3

N N
B B

Browser


Browser p1 p2 p3

N N
B B

Browser “That took a lot of work - and money!”


p1 p2 p3
“Not so great now…”
Web Tier B/L Tier
N N
B B
Web Tier B/L Tier

“That took a lot of work - and money!”
“Hmmm... Most of this stuff is sitting idle...”


Datacenter peak load

Usage Idle time

Jan Apr Jul Oct


“Wow! What a great site!”

Request
Browser Web Role Worker Role Azure
Storage
Response


Browser

Browser

Browser Web Role Worker Role Azure
Storage
“Server Busy”
Browser

Browser


Browser

Browser Web Role
N
Browser L Web Role Worker Role Azure
B Storage
Browser Web Role

Browser

You don’t see
this bit


Browser

Browser Web Role Worker Role
N N
Browser L Web Role L Worker Role Azure
B B Storage

Browser

OK, so changing config is easy...

…but what else do I need to know?


Browser

N N
Browser L Web Role L Worker Role Azure
B B Storage

Browser Tight coupling between web and
worker role is an anti-pattern


4. Process

Web Role Worker Role

2. Respond to UI

Req
1 Place on queue
. 3. Pickup from queue
Queue


4. Process


Req 5. Add result to table

Queue

6. Query table on refresh
Table

How many web and worker roles do you need?


WMI_Win32_PerfFormattedData_ASPNET_ASPNETApplications
/RequestsPerSecond
Browser

Browser Web Role
N
Browser L Web Role Worker Role Azure
B Storage
Browser Web Role

Browser

Management API
Log Worker Role
or Email


Web Role

Web Role


Req Req Req

Queue

Web Role

Web Role Monitor queue
length against user’s
Web Role expectations





Req Req Req

Queue


Web Role Monitor queue Worker Role
length against user’s
Web Role expectations Worker Role

Takeaways

A core tenet of cloud computing is
the ability to scale up/down

Understand how to communicate
between roles and nodes

Strategy on when to scale up/down
roles in production

Patterns for Moving to the Cloud

#2 - Using the Cloud for Multi Tenancy


“I like the idea of scaling Web roles…”


“…but need to serve multiple customers”


“Without creating separate codebases!”


Customer #1

Customer #2

Customer #3


Browser
Customer #1

Customer #2

Browser
Customer #3

Schema Customizations

UI Customizations

3 options for data in multi tenant environment


Browser
Fixed Schema
Customer #1

Customer #2

Browser
Option 1: Everyone Uses the Same Database/Schema
Customer #3
Pros: Simplest approach, easy to maintain/upgrade.
Cons: No customizations. Restoring of tenant data.


Browser Database
Customer #1 Customer #1


Browser Database

Option 2: Give Each Customer Their Own Database/Schema
Pros: Flexible. Tenant restore is easier. High Isolation.
Cons: Can be costly. Difficult to upgrade db schemas.


Browser Fixed Schema
with Customizations
Customer #1

Customer #2

Browser
Option 3: Fixed Database/Schema, with customizations
Customer #3
Pros: Customers can add their own custom fields
Cons: Non standard way of customizing the schema.
Tenant restore is difficult.

Tenant

*
1
TenantID int

TenantName nvarchar

*
Employee
EmployeeId int TypeID
*
Type
int

FirstName nvarchar Name nvarchar

LastName nvarchar CLRType nvarchar

1 1
Customization

*
int

*
CustomizationID

Value nvarchar

Tenant

*
1
TenantID 1

TenantName UK Branch

*
Employee
EmployeeId 1 TypeID
*
Type
1

FirstName Gordon Name PostalCode

LastName Brown CLRType string

1 1
Customization

*
1

*
CustomizationID

Value SW1 2AA
A

Tenant

*
1
TenantID 2

TenantName US Branch

*
Employee
EmployeeId 2 TypeID
*
Type
2

FirstName Barack Name ZipCode

LastName Obama CLRType string

1 1
Customization

*
2

*
CustomizationID

Value 20500

Tenant

*
1
Fixed Schema TenantID 2
Applies to All Could be both
TenantName US Branch

* *
Tenants

Employee Type
EmployeeId 2 TypeID 2
Customizable
FirstName Barack Tenant by Tenant Name ZipCode

LastName Obama Basis CLRType string

1 1
Customization

*
2

*
CustomizationID

Value 20500


http://hr.contoso.co.uk
Browser
Customer #1 http://employeedata.cloudapp.net
Customer #2

Browser
Customer #3
http://hr.fabrikam.com


http://hr.contoso.co.uk -> CNAME (employeedata.cloudapp.net)
Browser
Customer #2

Browser
Customer #3
http://hr.fabrikam.com -> CNAME (employeedata.cloudapp.net)


http://hr.contoso.co.uk -> CNAME (employeedata.cloudapp.net)
Browser
Customer #2 Use custom route to controller mapping

Browser
Customer #3
http://hr.fabrikam.com -> CNAME (employeedata.cloudapp.net)

Demo: Multi Tenant Schema and UI

Takeaways

Always consider Multi Tenancy first,
even if only one customer

Design considerations must include
both data and UI

Many other considerations, such as
identity – p&p guidance


#3 - Using the Cloud for Compute


Jim sees how cloud computing supports
scaling up/down nodes


“Can I use all of these nodes in parallel?”


“I’ve got this complex calculation I would like to
share across these multiple nodes…”


Data
Data Worker
Data
Client Master Worker

Data
Worker


Result
Worker

Result

Result
Worker
Result

Popularized by the term “MapReduce”*

* 2004 OSDI paper by Jeff Dean and Sanjay Ghemawat (Google)


How many ”e”’s in
“The quick brown fox jumps over the lazy dog”?
M<k,v>
Data Worker
M<k,v>

M<k.v>
Worker

“Map”

M<“the quick brown”.”e”>
M<“fox jumps over”.”e”>
M<“the lazy dog”.”e”>


L<k,v>
Worker

L<k,v>

Result
Worker
L<k,v>
3 x ”e”s found
“Reduce”
L<”e”,”1”>
L<”e”,”1”>
L<”e”,”1”>

And it’s definitely popular...


Google
20pb of data analyzed every day using MapReduce

Yahoo!
10k+ cores, 4pb of data using MapReduce

Facebook
2500+ cores, 1pb of data using MapReduce


Apache Hadoop
Open Source Java “Inspired by MapReduce”
(Core, HDFS, many more)

Cloudera
Consulting, training, distribution of Hadoop

Amazon Elastic MapReduce
Hadoop implementation on EC2


Data
Data Worker
Data

Client Application Job/Task Scheduler Worker
Data

Remote Service

How about implementing this on Windows Azure?


Data
Worker
Data

Data
Worker
Client Web Role
Data
Client Application Job/Task Scheduler Worker
Remote Service

Req

Queue

Table

Demo: Windows Azure Demo
“Inspired by MapReduce”

Takeaways

MapReduce very visible, although
can be difficult to initially grasp

Learn about existing frameworks,
especially Apache Hadoop

Read up on Dryad (DryadLINQ) for
future direction


#4 - Using the Cloud for Storage


“The cloud lets me store infinite data, right?”


Lots of headaches with data management today


“It sounds too good to be true…”

How does Jim do this today on premises?


File
RPC/NFS
Client File Server Store


Data TDS (Tabular Data Stream)
Client DB Server RDBMS


Data
HTTP POST
Browser Web Server Database


I love I love
you! you too!

Server Data

Affinity between your data and physical
hardware that serves it


I love I love
you! you too!

Server Data

Symptoms:
Which RAID number was that again?
Tedious to backup exponentially growing data
Crap! I’m at 95% capacity - got to move to a bigger disk

Breaks the affinity between your data and hardware


PutBlob Blob Container
PUT http://account.blob.core.windows./net/containername/blobname

REST API
Client Azure
Blob Storage

http://account.blob.core.windows.net/containername/blobname

PutBlob = 64Mb MAX
MetaData = 8Kb per Blob


Blob Container

REST API
Client Azure
Blob Storage
GetBlob
GET http://account.blob.core.windows./net/containername/blobname



Blob Container

REST API
Client Azure
Blob Storage
GetBlob
GET http://account.blob.core.windows./net/containername/blobname
Range: bytes=329300 - 730000


PutBlock(blobname, blockid, data) Blob Container
PutBlockList(blobname, blockid1 …, blockidN)
,
REST API
Client Azure
Blob Storage


PutBlock = 4Mb MAX to a maximum of 50Gb
BlockId = 64 bytes

REST:
GET http://account.table.core.windows.net/Customer?$filter=%20PartitionKey%20eq%20value
LINQ:
var customers = from o in context.CreateQuery<customer>(“Customer”) where o.PartitionKey
== value select o;

Worker Role Azure
Table Storage

http://account.table.core.windows.net
Each Table:

PartitionKey (e.g. DocumentName) to ensure
scalability
RowKey (e.g. version number)
[fields] for data

SQL Server Data Services (MIX08)


This is what I’m doing on premises...

Data
TDS


So, this is what I would like to do in the cloud...

Data
TDS


So, this is what I would like to do...

Data
TDS
Client SQL Azure RDBMS


HTTP TDS
Browser Web Role SQL Azure RDBMS


HTTP Queue TDS
Browser Web Role Worker Role SQL Azure RDBMS


“The Data Center”

HTTP TDS
Browser Web Tier Bus. Logic SQL Server RDBMS


“The Cloud”

HTTP Queue TDS
Browser Web Role Worker Role SQL Azure RDBMS

Takeaways

Storage in the cloud may look the
same, but breaks the affinity issue

Understand the pricing model for
storage on-premises vs. cloud

SQL Azure as a factor for
migration/move from on premises


#5 - Using the Cloud for Communications


Jim’s organization needs to communicate with
other organizations


“This has always been a very tricky and expensive
process to get working”


“Does the cloud offer anything to help?”

How would Jim have done this before?


Telco provided WAN
Company 1 Company 2

Client Server

FTP FTP
Client Server


Internet
Telco provided WAN
Company 1 Company 2

Client Server

FTP FTP
Client Server


Internet
Telco provided WAN
Company 1 Company 2
DMZ

Client Server

Extranet
Browser
Site


http://app.queue.core.windows.net/

REST
Client Azure Queue

Company 1

Client

REST
Azure Queue

Company 2

Client

Company 1

Client

Primary Access Key
REST
Azure Queue

Company 2

Client

Primary Access Key

Company 1

Client

REST
Web Role Azure Queue

Company 2

Client

Proxy?
Company 1

Firewall
Client

REST

Company 2

Client
Firewall

Company 1

Firewall
Client

REST

Company 2 65.55.33.204

Client
Firewall
192.168.14.100

NAT?

Company 1

Firewall
Client

REST

Company 2 65.55.33.204

Client
Firewall
192.168.14.100
OK, so I’ll do a HTTP
poll every minute

Company 1

Firewall
Client

REST

Company 2 65.55.33.204

Client
Firewall
192.168.14.100
…and how about other
protocols other than HTTP?

How does the .NET Service Bus help?

Company 1

sb://.../myqueue
Client

.NET
Service Bus

Company 2

Client

Two modes: TCP Relay and Message Buffer

Company 1

sb://.../myqueue
Client

.NET
Service Bus

Company 2

Client
Firewall
Outbound bi-directional socket
Kept alive in background

Company 1

Firewall
sb://.../myqueue
Client Outbound one-way socket

.NET
Service Bus

Company 2

Message routed
Client
accordingly
Firewall

This Conference
Application Firewall
sb://.../myqueue
My Laptop Outbound one-way socket
Localhost:1000

.NET
Service Bus

My Home

Message routed
Home PC
accordingly
192.168.14.193 Firewall
SQL
Express

Port 1433

Company 1

sb://.../myqueue
Client

.NET
Service Bus

Company 2
ATOMPub via REST

Client
Firewall
Consumer defines and creates
a Message Buffer

Company 1
Producer sends messages to
the Message Buffer
sb://.../myqueue
Client

Firewall
.NET
Service Bus

Company 2
ATOMPub via REST

Client
Firewall

Company 1

sb://.../myqueue
Client

Firewall
.NET
Service Bus

Company 2
ATOMPub via REST

Client
Firewall
Consumer picks up these messages
(retrieve, peek, lock)

Takeaways

Be careful consuming REST based
queues using shared secret

Firewalls/NATs can add additional
trouble (especially non-HTTP)

Learn how .NET Service Bus
traverses in these scenarios


Putting the Patterns Together


We’ve covered 5 patterns, but Jim has one last
question…


Are there cases where these patterns work
together?

Windows Azure
PHP

Let’s build an
application in PHP
that scales to many
nodes…

Use the principles of Windows Azure
multi tenancy to PHP (10 nodes)
create a version of
the application across
multiple vendors
Google AppEngine
Java Servlet

Amazon EC2
(5 nodes
ASP
.NET
using ELB)

Windows Azure
PHP

Job
Google AppEngine
Client Job Java Servlet
Compute results with
Job a MapReduce-like way
of distributing work
Amazon EC2 across all of these
ASP
.NET applications

Store the results
in a SQL Azure Windows Azure

database PHP

Job
Google AppEngine
Client Job Java Servlet SQL Azure
Database
Job

Amazon EC2
ASP
.NET

.NET Service Bus
Windows Azure
PHP

Job
Google AppEngine
Database
Job

Coordinate Amazon EC2
communication ASP
.NET
through the .NET
Service Bus

.NET Service Bus
Windows Azure
PHP

Job
Google AppEngine
Database
Job

Amazon EC2
ASP
.NET

.NET Service Bus
Windows Azure Store results
How many prime in SQL Azure
numbers between PHP
1 and 10,000,000? Compute

Job
Google AppEngine
Database
Compute
Job

40 jobs of
Amazon EC2
250,000
numbers ASP
.NET
Compute

Next Steps

Obviously, our friend Jim, is fake…

Next Steps

Make sure you have a clear definition
of Cloud Computing

Explore the 5 usage patterns for
your scenarios today

Think about the next steps for
implementation and migration

simon.guest@microsoft.com
http://simonguest.com

Patterns for Cloud Computing

Recommandé

Recommandé

Contenu connexe

Tendances

Tendances (20)

En vedette

En vedette (11)

Similaire à Patterns for Cloud Computing

Similaire à Patterns for Cloud Computing (20)

Plus de Simon Guest

Plus de Simon Guest (20)

Dernier

Dernier (20)

Patterns for Cloud Computing