SSDT is great for many things, but one thing it’s not great at is giving control over to the user with regards to data and log file creation. It uses prefixed values of DefaultDataPath and DefaultLogPath. These variables are system defined and cannot be overridden. Instead it uses the default locations that are set on the instance we are deploying to. Humph. And I’d wager that if these are null then it will just use where the system database settings are. This is not particularly useful for anyone.
You may think “Well OK I’ll just use a pre-deployment script”. Nice try. I thought so too. But as it turns out pre-deployment scripts aren’t pre-deplyoment scripts per se. If you add a pre-deplyoment script to create a database, it’ll be run AFTER the initial database creation (go on try it yourself if you don’t believe me, but I have saved you the trouble already!)
Clearly, we have to create a shell database that we can control where files are created outside of SSDT, and also the primary filegroup. If you’re using something like Octopus Deploy or TeamCity, or VSTS then it’s simple enough to create a step that will create this shell database for you. Then everything else (logins, users, tables etc) can be created using SSDT. And if you have more files/filegroups to be created, then you can use SSDT to set these up; what I’m talking about is the initial creation of the database.
There’s no shortage of options as to how to do this: you could have a SQL file with the CREATE statements run, and this would work, providing you have the correct “if not exists create db” setup, but this may not be particularly useful for re-use.
If you want to go for something that can be re-used, then I think SMO is your friend. It can be downloaded via Feature Pack, so practically any box can have it deployed (and what I mean by this is that you don’t need a full SQL Server install to have access to SMO), it works well with PowerShell, and there’s no shortage of examples online with how you can create a script that does what you need to do. This post from Allen White gives a good overview of SMO in PowerShell. And if you’re using something like Octopus, then this script also helps give a good start on creating a database using SMO. Where I think this step template in the Octopus library falls short is that it only creates the database, and not sets the files and filegroups up. But if you combine Allen’s scripts with the idea of the step template, then you can pretty much see where I am coming from. It’s also worth setting the recovery model using SMO and not worrying about this using SSDT.
I’m not wild about the use of scripts outside of the database to deploy the database; it implies “magic”, or behind the scenes changes that some people won’t know exists. And usually I’d advocate the use of SSDT pre/post deploy scripts to manage everything within the database. But databases are hard to deploy: they’re not a bunch of dlls that can be rolled back. A minor error early in the setup of a deployment pipeline can cause massive problems further down the process. And file creation is the first thing you do. What if the log file is created on the C drive? How hard can it be to move log files? Very hard is how. And creating a secondary log file to compensate is awful. And to try to remove that first log file? All these issues could be removed with a bit of forethought.
Remember it’s no good assuming that the instance you are deploying to has everything set up the way you like it. You have the control over this stuff when you are using CI/CD properly, and you have to take responsibility for database deployments.Follow @rPh0enix