Oslo – It’s a Floor Wax and a Dessert Topping…

2987577318_11aa449e20_m

At this year’s PDC, Microsoft has finally unveiled a first look at their new modeling platform, code-named “Oslo”.

The first question that comes to mind before we can talk about Oslo is “What is modeling?”

Modeling in software development is nothing new. In fact it has been attempted in several forms before. In the past, software modeling was done with CASE tools that attempted to turn UML type graphical structures into code. In general, these tools were very expensive and for the most part were tied to specific software development paradigms created by the thought leaders of that time (i.e. Booch, Rumbaugh, Jacobson, etc.) One of the more popular tools, Rational Rose, allowed users to create UML diagrams which could then be used to generate code.

While all of these various modeling technologies captured the imaginations of developers at that time, they really failed to deliver on the promise of making the software development process more manageable and understandable. I remember trying out Rational Rose and bumping up against the following issues:

  • The impedance mismatch between general software modeling language like UML and the underlying software languages of that time (i.e. C and C++). It was sometimes difficult to express in UML specific C++ constructs and vise versa..
  • The graphical nature of the modeling along with the monitor sizes of that time made it difficult to get a good view of complex software solutions.
  • Also, software, in general, was much more tightly coupled in those days making it more difficult to model as well.

While I’m sure there are success stories that can be cited from this style of modeling, it hasn’t been part of what I would call mainstream development. These tools were just too difficult and expensive making it questionable whether they provided any real advantage.

When I originally was exposed to these tools, I was writing shrink-wrapped software that was written for a single PC. Since this time software development has gotten much more complex. Throughout the history of software development there have been new abstractions introduced to make things easier so that we could produce more robust software solutions. For example:

  • Assembly Language was introduced to make machine code easier
  • C made it easier to create more complex solutions that were difficult in assembly
  • SQL provided a common language that could operate on top of structured data
  • C++/SmallTalk and other Object Oriented languages helped reduce complexity by allowing encapsulation into objects
  • COM/DCOM/CORBA was introduced to make it easier to break software into reusable components
  • NET/Java and other runtime languages provided an abstraction on top of the underlying operating system
  • XAML provides a more declarative language abstraction that describes what is desired rather than the imperative steps to accomplish the outcome

Often these technologies were received with mixed feelings. Some could see their potential value while others talked about the loss of control and performance that the abstraction would bring. But with many of these, the increase in computing power and the new solutions that were made possible justified the move to greater and greater abstraction.

I know that speaking from my own personal experience; the software development process has gotten more and more complex especially with regard to large enterprise solutions which have almost become unmanageable. Now with SOA-type architectures we no longer have systems that are deployed and run in a monolithic fashion. Instead these systems are an interconnecting maze of services, not all of which are under your direct control. Systems have gotten very difficult to maintain and configure. It also seems that software has a much longer lifetime making it very difficult to understand its intent especially as new developers come on broad and the experienced developers that originally authored the code leave.

Microsoft realizes that this complexity is just going to increase as we look forward to the future where some software will be deployed both on-premises and in the “Cloud” as well as to many other devices. These future solutions and experiences will be very difficult to create and manage using the tools we have today. Raising the level of abstraction to the next level will be needed, which is what Microsoft is attempting to do with the introduction of “Oslo” at PDC08.

Oslo is quite an expansive offering from Microsoft that brings Modeling into the forefront. Instead of just modeling the software itself, Oslo has the much broader and ambitious goal of modeling the entire software development lifecycle, from inception and business analysis to design and implementation to deployment and maintenance. With Oslo, these models are living and breathing. They not only organize and support the development process, they are actually used to run the software and systems using runtimes that consume these models.

During the PDC08, Chris Anderson and Giovanni Della-Libera quipped that Oslo was a “…dessert topping and a floor wax…” (click here for SNL reference). This overarching view where Oslo is fully integrated into the software development process is what makes this project so ambitious.

In short, Oslo is composed of three primary building blocks:

  • Repository: In Oslo all of the modeling data is stored in a Sql Server repository. This data contains all of the entities being modeled. This modeling data is comprised of such things as: software entities, workflows, business analysis, hardware, stake holders, etc.
  • Modeling Language (“M”): Oslo comes with a new modeling language called “M”. This language is has several dialects:
    • M-Schema : This dialect is essentially a Domain Specific Language (DSL) for modeling storage, more specifically database storage. Using this DSL, one can easily create a database schema.
    • M-Graph : This dialect is a DSL for adding data to a database. This DSL enables a very simple way to populate a database.
    • M-Grammer : This is the most interesting of the three because it is essentially a DSL for creating DSLs. Creating languages is very complex and previously left for the realm of researchers. Using M-Grammer makes it much easier to create textual Domain Specific Languages that allow users a way to code in specific domains.
  • Modeling Tool “Quadrant” : To be able to consume, browse, create and change the vast amount of modeling data associated with an enterprise software solution, it was necessary for Microsoft to create a tool that would make this feasible, this tool is code-named “Quadrant”.  “Quandrant” is essentially a graphical front end to the Oslo Repository.

Microsoft showed off all three of these components of Oslo at the PDC in the various presentations and keynotes. It was very obvious that we were seeing the very beginnings of the last few years of effort. Many of the demos that were shown were fairly simplistic making it clear that there is still a considerable amount of work to do before Oslo is ready for prime time.

At the PDC there were five sessions devoted to Oslo. These were all very informative and very well attended. I was able to attend many of these live and the rest I have watched online. Here are the links along with a quick summary of each.

A Lap around “Oslo”

Presenters: Douglas Purdy, Vijaye Raji

WMV-HQ WMV MP4 PPTX

This presentation was a great introduction to Oslo touching on all of its components. If you have time to watch only one of these sessions, I would recommend this one as it gives you a general feel for the potential of this new modeling platform.

“Oslo”: The Language

Presenters: Don Box, David Langworthy

WMV-HQ WMV PPTX

This presentation discussed the details of the “M” language itself. They showed how this language worked to generate schemas, data and briefly touched on creating DSLs. They showed the “Intellipad” tool that was used to edit and create “M”. I’m sure that this Intellipad functionality will at some point be integrated directly into Visual Studio.

“Oslo”: Building Textual DSLs

Presenters: Chris Anderson, Giovanni Della-Libera

WMV-HQ WMV MP4 PPTX

This was a very interesting presentation that showed how to create new textual domain specific languages (DSLs) using “M”.  Language creation can be a very daunting task. I’m sure anyone that that studied computer science in college will remember how difficult it was to use LEX and YACC to create a language. “M-Grammer”, which is what this session discusses, makes it very easy to create DSLs. I believe that DSLs will be more and more prevalent in the future and will enable applications to be created quicker and more reliable.  Many of us deal with DSLs in our day to day work already. These common DSLs include XSL, (a language for transforming XML), SQL (a language for querying data), HTML (a language for creating web content), etc.

I highly recommend this presentation. To get the most out of it, you may want to watch the two previous ones I mentioned before watching this one.

“Oslo”: Customizing and Extending the Visual Design

Presenters: Don Box, Florian Voss

WMV-HQ WMV MP4 PPTX

This presentation drilled into the “Quadrant” tool used to explore the models stored within an Oslo repository. While it was obvious that this tool is a work in progress and still has quite a ways to go to achieve its goals, you can get an idea about how this tool makes it easy to view the enterprise in a very easy and ad-hoc way. This session goes into the extensibility of the “Quadrant” tool and how it can be tailored to fit very different types of users.

“Oslo”: Repository and Models

Presenters: Chris Sells, Martin Gudgin

WMV-HQ WMV PPTX

This presentation gets into the repository itself. The repository is built on top of SQL server and provides secure access to the model data stored there. This presentation touches on the many models (written in “M”) that are included with the repository. These existing models are used to model things like: Identity, Applications, Transactions, Workflow, Hosting, Security, and Messaging (just to name a few). Chris Sells goes into showing how models are compiled and loaded into the repository and well as how to access model data from the repository. In addition to this he also covers the core services provided by the repository, which are: Deployment, Catalog, Security and Versioning.

Other interesting posts on Oslo:

In addition to these videos, here are some other interesting articles that I have stumbled across on Oslo:

Introducing “Oslo”

Why “Oslo”?

Martin Fowler on Oslo

Advertisements

No comments yet

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: