I think we can all agree – and if not, please let me know why – that the Entity Framework provides some awesome capabilities – mapping your conceptual schema to your data schema, isolation from the relational database and database schema, change tracking and identity resolution, full query comprehension and optimization, and more.  Yet, despite these features, it is difficult to figure out how and where to include the Entity Framework in your application architecture.

In preparation for this talk, I spent a lot of time looking at different strategies and architectures for using your data access layer (DAL) and the Entity Framework – or any O/RM tool, for that matter.  In the end, I settled on three approaches – certainly there are other patterns and approaches, but I found these to be the most relevant:

1. Entity Framework as the DAL
2. Full encapsulation of the Entity Framework
3. Partial encapsulation of the Entity Framework

Let me spell it out in a little more detail, as well as highlight what I feel are the different pro’s and con’s.

Entity Framework as the Data Access Layer

With this approach, you opt to use the Entity Framework as your data access layer.  This is a perfectly legitimate approach, and comes with a host of benefits in many situations.  Here’s a picture of the architecture:

As you can see, in this architecture aspects of the Entity Framework – like entities and the entity framework object context – are scattered throughout the various tiers.  This approach has many benefits, but it also introduces some challenges.

- Pro’s:

• Excellent isolation from the DB schema
• Independence from the relational database
• Object services, namely identity and change tracking
• Full query comprehension and optimization

- Con’s:

• In .NET 3.5 SP1 the “object first” approach is not well supported (updated in EF 4.0)
• Difficult to switch your O/RM, as you are tightly coupled to the Entity Framework
• Limited support for data validation

If you aren’t offended by the scattering of your entities and entity contexts throughout your application tiers, then this is a perfectly valid architecture.  That said, most people probably recognize that technology changes, and an application that works perfectly well with Entity Framework today may need to move to a newer technology in the future.  Consequently, you may want to look at an approach that encapsulates the Entity Framework, so that you separate your concerns and have high cohesion in your tiers without tight coupling.

Full Encapsulation of the Entity Framework

There are many benefits to encapsulation.  We have talked and read about ensuring a separation of concerns – e.g. high cohesion, loose coupling – for years now, and for good reason.  Typically we want to reduce the tight coupling throughout application so that tools like Entity Framework don’t proliferate through all the tiers of our application.  This helps make our code more testable, and allows us to utilize techniques such as Inversion of Control and Dependency Injection.

At the same time, fully encapsulating a tool like Entity Framework also comes with some costs.  If the real value of this tool is the ability to empower developers to quickly build powerful applications that are fully optimized against your database, then full encapsulation may inhibit some of the capabilities.

- Pro’s:

• Simplified queries
• POCO objects
• ORM & database agnostic
• No references to Entity Framework
• Isolation of SQL queries

- Con’s:

• Loss of change tracking and identity management with Entity Framework
• Additional object materialization of business objects
• No query composition; simple LINQ enumeration queries

Again, this is not a full laundry list, but hopefully it provides some of the different benefits and challenges.  Fortunately, there’s a third path you can take, which is a hybrid of the two – I call this partial encapsulation.

Partial Encapsulation of the Entity Framework

When you decide to partially encapsulate the Entity Framework, you acknowledge the benefits of the separation of concerns and do what you can to loosely couple your tiers.  But you also try to preserve the advantages of a tool like the Entity Framework.  In this scenario, you might expose your entities to the different tiers of your application, but encapsulate the rest of the Entity Framework in your DAL.  When you return data from the DAL, you would simply use one of the three techniques: IEnumerable, IQueryable, or ObjectQuery.  Other than the last technique, you aren’t exposing aspects of the Entity Framework – other than the entities themselves – to any other tiers.  This preserves the loose coupling, but allows you to leverage the real value of the Entity Framework.

- Pro’s:

• Query composition
• Identity resolution & changing tracking
• Some independence from O/RM tool

- Con’s:

• Use of Entity Framework entities has implications on business layer
• Lack of isolation of SQL queries – spread throughout the various tiers

Partial encapsulation is a pragmatic approach to this problem, and it’s a technique the I have seen many customers and partners utilize very successfully.  In the end, the approach you choose depends on your situation.  All three of these approaches are perfectly valid, but aren’t necessarily valid in every situation.

Below are the slides I used during the presentation.  I am sure that this is a presentation I’ll give again in the future, so expect to see updates and/or changes.

I’d love to hear your feedback, so please leave me a comment or message.

I hope this helps!

I recently learned that Statera has been investing a lot of time and energy in developing some cool applications using our Azure Services Platform.  In particular, Statera built an engine running in Windows Azure that integrations Microsoft CRM Online and SharePoint Online (both of which are part of Microsoft’s Business Productivity Online Suite, or BPOS).  It’s a great example of how you can use Microsoft’s cloud to extend existing applications.

This engine is called Stratus – love the name! – and you can find out more about it at http://stratus.statera.com/.

Here’s a quick laundry list of some of the features/capabilities:

• CRM users can provision SharePoint sites for Account, Opportunity and Case records, allowing them to store documents in a location where they are easier to collaborate on and in an environment where non-sales employees may be more familiar with.
• CRM users can upload documents while in CRM and they are stored directly in SharePoint.
• CRM users can view associated SharePoint document libraries and their contents and pull up documents without ever leaving CRM.
• SharePoint users can navigate to SharePoint sites that are associated with CRM Account, Opportunity or Case records and access uploaded documents, CRM contacts, CRM activities as well as some other general CRM information.
• SharePoint users can create new contacts or activities via the associated SharePoint site that are in turn stored in CRM.
• SharePoint users can save CRM contacts directly into their Outlook contacts folder with the click of the mouse.

Evidently, you can also use the engine to link your on-premises versions of SharePoint and CRM to BPOS, and any combination thereof.  Currently, it doesn’t appear that the on-premise activation is available.

Watch these guys, as I’m sure they’ll have other cool things out in the future.  For more information, check out their webcast http://stratus.statera.com/.

Ultimately, my goal in this talk was to address the following:

• Provide an introduction to software architecture – i.e. the practice of architecture, the role of an architect, architecture patterns & anti-patterns, methodologies, and the like – to developers that typically live in a world of software implementation.
• Explore the practical aspects of applying architectural practices in .NET.

I had a lot of fun preparing for this talk, and I sincerely hope that the audience found it valuable.  I hope to eventually record this as a webcast and then break off into more specific topics that dig into different areas.

The first – and only – book on Commerce Server 2007 is now available! (Here’s the translated version, thanks to Microsoft Translator - http://bit.ly/Gfm7R; note that you cannot purchase the book when using translator.)

Unfortunately, the only location you can currently purchase the book is from Amazon.de – not sure why it’s not available at Amazon.com, but hopefully this gets resolved soon.  I can assure you that this is an English version, and not German.

That aside, I really hope you find this a useful book.  It’s been a longtime coming, but it’s here now!  Go buy it!  If you have any issues purchasing this book, please let me know ASAP.

One of the major announcements was new capabilities for the Windows Azure CTP. This includes the support for FastCGI, Full Trust, and Geo-location. Please take a look at the Windows Azure team’s blog for all the announcements.

In this post, I want to walk you through the step you need to take to run a PHP application on Windows Azure. As you’ll see, it’s quite simple and straightforward – after a few configuration updates and additional steps, you can run nearly any PHP application. This walkthrough assumes that you have installed the latest bits for Windows Azure. If you are just getting started with Azure, please visit http://www.azure.com/ – this is the primary landing page for the Azure Services Platform. If you want to get the updated bits that allows for Full Trust and FastCGI, you can grab the bits at the Windows Azure Tools for Microsoft Visual Studio March 2009 CTP page.

1. Create a new project. Select File –> New –> Project.
2. Select Cloud Service project type and select the Blank Cloud Service template. Specify the name/location, and click OK.
   
3. Add a Cgi Web Role project. Right-click the project (e.g. CloudService1) and select New Web Role Project.
   
4. Specify the Name and click OK.
   
5. By default the Web.roleconfig file is opened in the new project. Remove the comments and update the application fullPath to “%RoleRoot%\php\php-cgi.exe”.
   
6. Now we need to actually grab the PHP executable that will bet invoked. Browse to http://www.php.net/downloads.php and grab the non-thread-safe Windows binary (at the time of writing, it is “PHP 5.2.9-1 Non-thread-safe zip package”).
7. You might as well configure PHP to run on your Windows environment. Rather than specifying all these steps, please see Simon Guest’s post on hosting PHP applications in 5 easy steps. Follow the steps in his post, and at the end you should be able to confirm that PHP works in IIS 7.
8. Copy all the files in the \PHP folder you downloaded into your Cgi Web Role project. XCopy works really well. For example, open a command prompt in the root of your Cgi Web Role project and type:
   xcopy /s \php php
   When asked, specify “D for directory”. This will copy everything from c:\php into a folder called php in your Cgi Web Role project. Note that this is where we pointed Windows Azure to in step 5.
9. Select your Cgi Web Role project, show all files, and add the new php folder into your project. This way it will get packaged up with the rest of your application and deployed to Windows Azure.
10. Open up the Web.config file and specify a new default document. This will tell Windows Azure to open up our PHP page by default. Under the <system.webServer> element, add a <defaultDocument>:
    
11. Additionally, we need to setup a handler in the Web.config file. Add the following code to the <handlers> element in the web.Config:
    
12. In order for this to function, we have to enable native code execution in Windows Azure. To do so, open the ServiceDefinition.csdef file and set the enableNativeCodeExecutiion flag to true:
    
13. Now, let’s create a new PHP file. Right-click on your Cgi Web Role and add a new item. Select the Text File template and change the name to index.php. Click Add.
    
14. To test, create some PHP code. A quick, but effective, test is to simply echo “Hello World”. In the index.php file, write the following code:
    
15. Press F5, and your browser should open and show you the following page:
    

And that’s all it takes!

While this is a simple demonstration, I hope it impresses upon you the impact that allowing native code execution has in Windows Azure. With native code execution, and fast CGI support, you can now leverage all the rich content created in the PHP community on Windows Azure. The story even gets better with the new relational capabilities of SQL Data Services, but I’ll save that for another post.

I hope this helps!

Please take a look at an announcement recently made by Glen (Software) Smith announcing that the Commerce Server 2007 book is finished.  You may (or may not) recall that this book was originally going to be published by Wiley (aka Wrox), but that it was cancelled for a number of reasons.  Consequently, efforts on the book essentially stalled until Glen stepped up last fall and volunteered to see it through fruition.  I am extremely grateful for all his hard work.

Glen has done a fantastic job taking the existing content I created (along with input from Tom Schultz, Ryan Donovan, Kevin Kraus, Brad Bont, David Hargis – and even Max Akbar) and extend it with additional content.  Furthermore, I’m very excited that he’s currently refreshing the book so that it contains information on Commerce Server 2009 as well – very exciting!  Take a look at the TOC (note: this may change based on the refresh for Commerce Server 2009):

   i. Introduction
Part I: Commerce Server Primer   
   1 Getting Started
   2 The Commerce Server Architecture
Part II - Commerce Server Overview   
   3 The Catalog System
   4 The Inventory System
   5 The Orders System
   6 The Profiles System
   7 The Marketing System
   8 The Commerce Server Adapters for BizTalk Server
   9 Commerce Server Staging
   10 The Data Warehouse Analytics System
   11 The Business Processing Pipeline System
Part III: Developing with Commerce Server   
   12 Commerce Server Development
   13 Developing with the Catalog System
   14 Developing with the Inventory System
   15 Developing with the Orders System
   16 Developing with the Profile System
   17 Developing with the Marketing System
   18 The Business Management Applications
   19 Extending the Data Warehouse and SQL Reporting Services
   20 Commerce Server Integration
   21 Operational Management
Part IV: Commerce Server Production Architectures   
   22 Application Architectures for Production

I know you all have questions about the cost and availability of this book.  When I have more information, I will post it.  For now, if you have questions, please feel free to email me (through my blog here) or reach out to Glen at his blog (http://www.software-smith.com/).

Looking back at my first announcement regarding Commerce Serve 2007, I’m struck by the fact that it’s been almost two years – amazing.  What’s more amazing is that today there STILL isn’t a definitive book on building e-commerce applications with Commerce Server 2007 (or 2009, for that matter).  I think you all will be very happy with this book.

Hope this helps!

Bryce gave an excellent presentation on cloud computing that fostered some good discussion within the group.  His presentation was divided into the following topics:

• What’s it all about?
• Why take it seriously?
• Software + Services
• Azure: Microsoft’s Cloud Services Platform
• Developer Experience
• Business Models

Bryce also dropped into Visual Studio and built out a Windows Azure application that communicated with data stored in SQL Data Services - simple yet effective.

I was pleased when Bryce showed this quote from Steve Ballmer early in the presentation, as I think it really captures the value of engaging with this user group:

The Azure Services Platform was announced at the Microsoft Professional Developers Conference (PDC) this past October, and you can already start building applications today through the public CTP.  Now is the time to explore the technology, try out new and different ways of writing software, and provide feedback to Microsoft on what works and what doesn’t work!  It is my hope that this user group becomes a forum to explore how the tools and technologies that are available through Azure help (or hinder!) businesses when they attempt to create compelling applications in the cloud.

Below are some of my notes from the presentation.  This isn’t an exhaustive set of notes, but rather a few pieces from the presentation that I found interesting and compelling.  If you want a summary of the entire presentation, well, perhaps you should come to the next meeting.

Why take it seriously?

I really liked Bryce’s discussion on why we should all take the cloud seriously.  In particular, I felt that the following slide highlighted how all the big players - Microsoft, Google, Yahoo, Amazon, Oracle, etc. - are investing, to some degree, in the cloud.

Software + Services

Bryce had a slide that discussed S+S that I found rather interesting.  In this slide …

… you can see that, on the left, we have a fairly standard architecture for on-premises software.  You might find this in any enterprise data center, as it is specifically designed to handle the needs and scale of an enterprise.  On the right you can see services in the cloud - notice how the various tiers are designed to scale horizontally and provide “on-demand” scaling.  This architecture is significantly different from your standard on-premises architecture because you are now dealing with Internet-scale, something that most companies and enterprises have little experience at providing.

What’s missing from this picture, in my opinion, is an illustration that shows that S+S is not an either/or proposition.  Instead, S+S is about choice and flexibility.  There’s no expectation that businesses will move all their on-premises assets and applications into the cloud; instead, they will probably identity specific instances in which it may make sense to move an application, a database, or some other asset into the cloud, and then use technology (perhaps provided by Microsoft’s .NET Services) to bridge their on-premises systems to their cloud-based systems.  It’s all about choice.

I really like this slide from Ray Ozzie’s keynote presentation at PDC:

You can see here that the technologies provided by Microsoft provide a continuum that gives you an on-premises platform as well a set of cloud services that complement the on-premises platform.  You can use what makes sense and works in your scenario.

Business Models

The best source of information regarding the Azure business model (i.e. pricing and licensing) continues to be the Pricing & Licensing Overview page.  There you can read about the four principles of the business model:

1. Consumption-based model
2. Pricing attractive with the market
3. Market expansion opportunity for Microsoft partners
4. Easy access through the Web, or through existing channels and programs

Bryce made some good points regarding Opex (operational expenditure) and Capex (capital expenditure) when speaking to this slide:

This slide attempts to show that you can significantly reduce your costs when you no longer have to maintain a large infrastructure to support your on-premises software.  Instead, if you move towards a utility compute model, you only pay for the services you need and let the provider absorb the cost of maintaining the data center.

I noticed a little healthy skepticism from the group, and rightly so.  This is a new platform, and the onus is on Microsoft to show that they (er, we) can deliver on the promise of cloud computing.  Some of the top concerns I heard from participants included:

• The security provided by Azure storage and SQL Data Services
• The (perceived) disconnect between SQL Server and SQL Data Services and confusion regarding the roadmap
• The business model and lack of specific pricing and licensing details

It seems to me that these topics would make excellent topics for future posts, and I will try to do so.  In the meantime, I encourage those of you that have these same concerns and questions to engage in your community, try out the new technology, and explore the capabilities of the Azure Services Platform - all you have to do is go to http://www.azure.com/ to get started.

All in all, I am very excited about this new users group and the community it will create, and I appreciate Neudesic’s efforts to launch these groups around the U.S (see http://cloudcomputingusergroup.com/ for details).  The current plans are to have another Chicago user group meeting in January, most likely held at the Microsoft office in downtown Chicago.  Once I hear something concrete I will be sure and let you know.

For those of you that attended the user group meeting (and even those of you that did not, for that matter), please feel free to reach out to me if you have any questions on the Azure Services Platform, and how you might start building some compelling applications.  You can reach me through this blog.

It was an incredible journey to PDC, and I plan to post a number of blog entries over the next couple weeks highlighting our day one keynote with Bob Muglia and RedPrairie (skip to minute 59:00), our use of the Azure Services Platform to build a supply chain application, and lots of code and patterns that show how we did it.  For now, I’d like to share with your our breakout session at PDC2008.  To watch our presentation, please click the picture below:

Our session abstract:

Learn how we built a multi-enterprise supply chain application using the cloud services platform. Today, most organizations are deeply interconnected, and business applications span multiple enterprises. See how we built an application that satisfies requirements around connectivity, identity, orchestration, and storage, providing a scalable, pervasive, highly available, general-purpose platform that replaces custom software and infrastructure. Hear about the core requirements for multi-enterprise business applications, their architecture, and the cloud-based framework we built to support them. Learn about the issues we encountered, lessons learned, and useful guidelines and patterns.

If you are interested in building out higher-level services on top of the Azure Services Platform, please let me know!  We are looking for great customers that want to explore the many possibilities that the cloud has to offer.

Enjoy!

For the last two months I’ve been working on a project that uses the “Windows Cloud OS” and “Cloud Services”, and I can’t wait until we disclose all the details at PDC 2008.  Expect to see a significant amount of details regarding this cloud computing project as we get closer and closer to PDC.

In the meantime, enjoy a sneak peak at the work done by one of our ad agencies – can you feel the excitement??

If you have any interest in the new Windows 7 and Microsoft’s cloud computing strategy, you won’t want to miss Microsoft PDC 2008!

A couple of notes:

• This assumes you’re running Windows Vista or Windows Server 2008
• I’ve tested by installing the following O/S’s: Windows XP SP3, Windows Vista, and Windows Server 2008
• You must have a machine capable of booting from a USB drive

Without further ado, here are the steps:

1. Open an elevated command prompt.
2. You must make your USB drive bootable.  Type the following in the command prompt:

   diskpart
   list disk (FIND YOUR USB DISK)
   select disk 1 (OR WHATEVER YOUR DISK NUMBER IS)
   clean
   create partition primary
   select partition 1
   active
   format fs=NTFS
   assign
   exit
   

3. Mount your Windows Server ISO (or unpack it) and copy the contents of the CD/DVD onto your USB.  Make sure you get all files, including hidden ones.

You should now be able to use this USB drive to install a new O/S.  Good luck!