It is no secret that Microsoft has been working on bringing its Enterprise offerings up to date, readying them for the next generation of applications and services, and fixing small pain points that have vexed developers for years. In just a short while, BizTalk Server 2006 R2 will make way for BizTalk Server 2009, and another interesting product from Microsoft will realize next version status. That product is Microsoft Enterprise Service Bus (ESB) Guidance 2.0.
What is an Enterprise Service Bus?
Dmitri Ossipov, a Senior Program Manager for Microsoft working on the ESB Guidance, in his interview on .NET Rocks defined ESB as an "architectural paradigm for policy driven mediation." Nicholas Allen, a Program Manager at Microsoft working on BizTalk Server, argues that "the clearest definition of what companies think ESB means comes from looking at the products that they build." In the case Microsoft's ESB offering we see a solid implementation of the Routing Slip pattern built on top of BizTalk Server sprinkled with ample extensibility points. In the world of the ESB Guidance Routing Slips are called Itineraries, and act like an order placed at a menu of services that is the bus. The ESB Guidance provides flexibility through a loosely coupled design that allows routing and transformation decisions to be made at runtime instead of having to be statically configured at design-time. This enables service composition, dynamic transformation, and adds support for scenarios previously unimaginable in a BizTalk Server environment.
Version 1.0 of the Guidance was a paradigm shift for many BizTalk and .NET developers, but version 2 has the potential to take it to the next level. It introduces some killer new features such as the Itinerary Designer that can reduce XML induced eye-strain, Generic On-Ramps that allow you to send a message into the bus on the consumer's terms, and support for Server-side Itineraries that can place ESB developers back in control of the content of their Itineraries.
Itinerary Designer
Someone once said, "XML is like violence, if it doesn't solve your problem, you're not using enough of it." I'm not going to debate the truth of this one way or another, but I do find it interesting that XML is compared to something that causes such a universal adverse reaction. When color-coded, perfectly indented, and collapsible, I can handle XML. However, at that point I have already resorted to looking at a more human friendly representation of the data instead of the raw data itself.
Those who have downloaded the January CTP of the Guidance, have found themselves in the midst of peace – no XML to be seen (don't worry it's still there if you dig). The January CTP now includes a Visual Studio designer for Itinerary models. Creating Itineraries is now as simple as dragging On-Ramps, Itinerary Services and Off-Ramps into a visual model that can be exported to a repository, or as XML – even mere mortals can do it.
Server-Side Itineraries
Yes, I just said the word repository. Version 1.0 of the guidance was awesome, but it did leave developers with a puzzle: "How do I get the itinerary I need to route this message, and how do I get it to the server?" The answer of course was that you have the XML of the itinerary that you send within the header of the message when submitted to the Itinerary Processing web service.
But where do you get the XML from? How do you know it's valid? Well, with CTP2, you know an itinerary is valid because it was modeled in a designer that validated it before each save and export. With CTP2 the XML can be retrieved from a SQL database (which bears a striking resemblance to the rules set database in BizTalk Server), and applied to the message in a pipeline component.
BizTalk and WCF enthusiast Bram Veldhoen remarked on his blog that it would "be a good idea to have the ESB be responsible for assigning the Itinerary headers." Microsoft apparently agreed, and this is exactly what should be expected from this new version of the ESB Guidance.
New Resolvers
The latest ESB 2.0 CTP adds three new resolvers for resolving itineraries: BRI, ITINERARY, and ITINERARY-STATIC. This means that not only can consumers rely on the ESB to apply itineraries for them; the ESB can do it dynamically. For the ESB Guidance uninitiated, Resolvers are these wonderful classes within the ESB Guidance that can take a configuration string, parse it and execute a query of some sort to look up information necessary for transformation, routing, or some other custom process.
The first is the BRI resolver; a typical resolver connection string would look like this:
BRI:\\policy=SamplePolicy;version=1.1;useMsg=True;
This string, when interpreted by the BRI resolver (the BRE moniker was already taken for a resolver that cannot retrieve itineraries from the repository), will tell the resolver to use the Business Rules policy named SamplePolicy to determine the Itinerary to use for routing. It will also include the message as a fact when calling the Rules.
The second is the ITINERARY resolver; a typical resolver connection string would look like this:
ITINERARY:\\name=Zebra;version=1.0;
This is a static choice of an itinerary named Zebra. Its sister resolver ITINERARY-STATIC does exactly the same thing but is implemented using the Unity Application Block, but that's a discussion to save for another posting.
You would use such resolver connection strings in the configuration of the ESB Itinerary Selector pipeline component, which is part of the ItinerarySelect* family of receive pipelines included with the ESB Guidance 2.0 CTP. Since this is all part of a pipeline, that means that in 2.0, you can create Itinerary On-Ramps that are first class citizens in the ESB which use transports other than an ASMX web service or WCF web service. The possibilities are limited only to the adapters installed.
Getting it Right
The new version of the ESB Guidance brings new features, enhancements, and fixes that really make it feel like a polished product. With version 1 they got it shipped, but with version 2 they're getting it right.
