Home > Articles > Data > SQL Server

SQL Server Reference Guide

Hosted by

The Utility Control Point and Data Application Component, Part 2

Last updated Mar 28, 2003.

This is part two of a tutorial in a series where I’m covering two new features introduced in SQL Server 2008 R2 Called the Utility Control Point (UCP) and the Data Access Component (DAC). In part one, I explained the UCP, since the two features work more or less hand-in-hand, meaning one is not as useful without the other. The UCP collects trending data for an Instance of SQL Server and — more importantly for this article — it can collect data from a DAC. In this tutorial I’ll show you what a DAC is, how you can create one, how to use it to deploy and alter databases for an application, and finally how you can use the UCP to monitor what you deploy with the DAC.

The Database Application Component

The Database Application Component (DAC) is comprised of two things: a ZIP file (with a .dacpac extension) and some entries in the MSDB database on a server. While those things are quite simple, creating and maintaining a database with it is more interesting. Once you’ve “deployed” a DAC (using the DACPAC) the UCP can monitor it, if the Instance is enrolled in the UCP. I’ll show you the entire process from beginning to end. You can follow along on a test system if you like, assuming you have at least two Instances: One of them set up as a UCP and the other enrolled as a monitored Instance. You can see how to do that in the first article in this series.

Setting up a DAC From Visual Studio 2010

I’ll start off with setting up the DAC using Visual Studio. This example uses Visual Studio 2010. I won’t spend a lot of time on creating the actual tables and other database objects, since there are other references for that, and I’ll point to those at the bottom of this article.

In the screen below you can see that Visual Studio 2010 has a new project type, called “SQL Server Data-tier Application”:

Once opened, you set the normal properties for a project, and then you can start to add database objects. The full list of supported DAC objects is here, and you should review that list carefully.

You can also add SQL Server 2008 (and higher) Policies — from Policy Based Management – that allow the developer to decide things like the operating system version, languages, even the Edition of SQL Server that they would like the application to reside on. This is VERY powerful — and allows the developer to communicate to the DBA team, literally blocking an installation of the database if it does not meet the criteria that the developer wants. In that way the developer can ensure that the best practices for that application are followed.

Once complete, the developer has two options — they can deploy the application directly to a server, or they can save the DACPAC out and allow the data professional to do it. I normally advocate the latter.

I’ll assume for this article that a simple application database, consisting of one table and one column, is created using this method. I’ll show you the T-SQL version of that in a moment — the point is that the DACPAC file is now on a hard drive somewhere, and that’s the contract the developer and the data professional will use to move the data around to each other.

Setting up a DAC from SQL Server Management Studio (SSMS)

While you have the option of creating the entire database offline in Visual Studio, you have other choices as well. In some shops the Data Professionals (DBA’s and others) create the database. You can still follow that process and work with a DAC — in fact, you can create a DAC right from SQL Server Management Studio (SSMS).

To do this, you’ll need to follow all of the restrictions I mentioned in the last Article. I’ve got two instances shown here, one called “UNIVAC” and the other called “UNIVAC\PRODUCTION”. You can see the databases on each of those systems listed below:

I’ll start this demonstration by creating a database with one table and one column using this simple bit of code, all against what I’ll use as “Development” in this situation, the UNIVAC server. Again, check the supported objects list above for a more comprehensive list of what you can include in an actual production system:

CREATE DATABASE DACTest001;
GO
USE DACTest001;
GO
CREATE TABLE TestTable001 (c1 int);
GO

Now I’ll right-click that database in SSMS and Tasks | Extract a Data-tier Application from the menu:

I’m shown a welcome screen, click the Next button, and then I’m asked to set the name, the version (more on this in a moment) and a description for the application. I’m also able to set the location for the DACPAC.

Clicking the Next button here checks to make sure all of the objects will handle a DAC. In this simple case, of course, I’m covered, but you have the ability to go change things here if this fails, leave this screen up while you do that, and then “re-run” the validation so that you don’t have to type everything back in again. It really isn’t that much info any, so I normally just close it and start over after I make my database modifications.

From there the system creates and saves the DAC in a DACPAC file. I always store that in source code control.

Deploying the DAC using the DACPAC

Regardless of how the DACPAC was created, you can deploy it from SSMS with the method I’ll use here. Recall that in Visual Studio you could have deployed the application directly from there, but you can also save it as a DACPAC.

I’ve moved to the UNIVAC\PRODUCTION system, and right-clicked the name of the Instance. Notice that this is a different location for working with a DAC since the database doesn’t exist yet.

I then select Deploy Data-tier Application... from the menu that appears. I won’t go through all of the screens here- it’s pretty much a “Next, Next, Finish” process, with one exception. At one point you’re able to change not only the name of the database but the location of the data files.

And in no time the system has created the database in UNIVAC\PRODUCTION, complete with a single table:

So far, you could have just backed up a database and restored it. The difference is that the scripts were run on this system – meaning that the users will be created properly, the Instance information stays and so on. Also, the database has now been registered in UNIVAC\PRODUCTION’s msdb system database — which means when the UCP collects data on the Instance the DAC will automatically be listed.

Migrating Changes with the DAC

No I need to make a change to the original database on UNIVAC. In this simple example, I’ll just add another table, like this:

CREATE TABLE TestTable002 (c1 int);
GO

Which shows up in the UNIVAC system (or over in Visual Studio if that’s where you’re working from) but not in UNIVAC\PRODUCTION. To move that change to the other system, I can use the DAC.

I follow exactly the same process as before, with one exception — I’ll change the version number:

Now, on UNIVAC\PRODUCTION, I move to the “Management” node in SSMS and locate the “Data-tier Applications” sub-node. I then simply right-click the DACTest001 (the name of the application I set) and select Upgrade:

Once again, this is kind of a “Next, Next, Finish” process. The wizard wanders through the DACPAC file and compares it to the schema it has on UNIVAC\PRODUCTION, and then makes the change:

There are some things to consider here. What the system actually does is rename the database, migrate the data in place, and then finally drop the original database. That means you’ll need at least twice the space to do one of these upgrades. It’s the safest way to perform the process, but also very expensive. This is why the DAC and UCP is best suited for smaller applications — performing this operation on a 10-terrabyte database is really not a good idea! I’d stick with databases below 5GB or so.

Monitoring the DAC with the Utility Control Point

After about 15-20 minutes, the Utility Control Point will begin to collect information on the DAC – and it’s the same information I explained about the Instance in the first series. But the key here is that you’re looking at the Database – not the entire Instance. That can be very helpful in finding out what your “higher end” applications are doing:

To recap, the DAC and UCP work in concert to bring you a trending view of the resources for your systems. They have quite a few restrictions, and are designed to monitor multiple, smaller apps on anywhere from 20-200 Instances to help you understand your load distribution from a CPU and Disk subsystem level.

InformIT Articles and Sample Chapters

Software migration and versioning is normally the domain of the developer — and you can find more about that in Application Deployment — Versioning from the InformIT .NET Reference Guide.

Books and eBooks

Here’s a very cool work on some of the newer thoughts in the design process, including upgrades: Emergent Design: The Evolutionary Nature of Professional Software Development.

Online Resources

There’s a full whitepaper on how a DAC works here.