How to track data changes when deploying a SQL SSDT project - sql

We upgrade out databases using SSDT DACPAC deployments. Developers work on the projects in VisualStudio 2015, modifying the schema where needed.
Developers also add Pre and Post deployment scripts to the projects. Some of these scripts ensure that certain tables always have the expected data in them. Others add, move, or mutate data as part of the platform upgrade.
We need to improve the output generated during database deployments so that, after a deployment, we have a human readable list of any data that changed as part of the deployment.
I'm currently considering two approaches. But neither seems ideal. They are:
1) Manually add logging to all Pre and Post scripts in the project. This is certainly an option. But it is not ideal because it complicates the upgrade scripts and is also likely to sometimes be missed or incorrectly done by developers. Since the goal is to detect unexpected data changes that occurred as part of the deployment, this uncertainty is a real bummer. A generic solution is preferable.
2) Here's my best stab at a generic solution: As part of the deployment process, enable SQL Change Data Capture for all user tables in the database. Then, at the end of the deployment, collect all captured changes and disable CDC. I've actually got this working. But the process of enabling CDC on all tables in the database takes a few minutes (one of our databases has 775 tables, it takes about 3 minutes to enable CDC on all of them). This approach also just feels very... heavy?
My question is. Is there a better way to achieve my goal of reliably generating a report of data that was changed as part of the database deployment, given that the deployment runs arbitrary Pre and Post deployment scripts?
If there does not seem to be a better way, I would appreciate feedback on option #2. Am I crazy for considering this?

Some of these scripts ensure that certain tables always have the expected data in them
First of all PreDeploy/PostDeploy scripts are not validated. I prefer to use them as launching point only, and do the actual work inside stored procedures.
So instead of writing:
INSERT INTO dbo.tab1(id, col1, col2) VALUES (...,..., ...);
and inserting it on PostDeployscript, you could put:
EXEC dbo.Populate_Tab1;
and define stored procedure as:
CREATE PROCEDURE dbo.Populate_Tab1
AS
BEGIN
-- idempotent script, here by using MERGE
WITH src(id, col1, col2,...) AS (
SELECT 1, ..., ... UNION ALL
SELECT ...
)
MERGE dbo.tab1 trg
USING src
ON trg.id = src.id
WHEN MATCHED THEN UPDATE
...
WHEN NOT MATCHED BY SOURCE THEN DELETE
WHEN NOT MATCHED BY TARGET THEN INSERT
...
END
Key point: stored procedure must be idempotent.
This way you are always sure that the table contain desired input data, and stored procedure is validated.
Similar approach is described by Kamil Nowinski:
Script and deploy the data for database from SSDT project
Advantages:
The stored procedures are part of the database project.
They will be validated and compiled, thus you avoid potential errors with uncontrolled code
The changes to the SP appears in output script only when something changes.
It’s easier to review the script before a run as it doesn’t contain unnecessary code
In Post-deployment script – there is only one line of code which never is changed.

Related

SQL (or any relational db) engine with SCM-friendly backing store [duplicate]

I'm doing a web app, and I need to make a branch for some major changes, the thing is, these changes require changes to the database schema, so I'd like to put the entire database under git as well.
How do I do that? is there a specific folder that I can keep under a git repository? How do I know which one? How can I be sure that I'm putting the right folder?
I need to be sure, because these changes are not backward compatible; I can't afford to screw up.
The database in my case is PostgreSQL
Edit:
Someone suggested taking backups and putting the backup file under version control instead of the database. To be honest, I find that really hard to swallow.
There has to be a better way.
Update:
OK, so there' no better way, but I'm still not quite convinced, so I will change the question a bit:
I'd like to put the entire database under version control, what database engine can I use so that I can put the actual database under version control instead of its dump?
Would sqlite be git-friendly?
Since this is only the development environment, I can choose whatever database I want.
Edit2:
What I really want is not to track my development history, but to be able to switch from my "new radical changes" branch to the "current stable branch" and be able for instance to fix some bugs/issues, etc, with the current stable branch. Such that when I switch branches, the database auto-magically becomes compatible with the branch I'm currently on.
I don't really care much about the actual data.
Take a database dump, and version control that instead. This way it is a flat text file.
Personally I suggest that you keep both a data dump, and a schema dump. This way using diff it becomes fairly easy to see what changed in the schema from revision to revision.
If you are making big changes, you should have a secondary database that you make the new schema changes to and not touch the old one since as you said you are making a branch.
I'm starting to think of a really simple solution, don't know why I didn't think of it before!!
Duplicate the database, (both the schema and the data).
In the branch for the new-major-changes, simply change the project configuration to use the new duplicate database.
This way I can switch branches without worrying about database schema changes.
EDIT:
By duplicate, I mean create another database with a different name (like my_db_2); not doing a dump or anything like that.
Use something like LiquiBase this lets you keep revision control of your Liquibase files. you can tag changes for production only, and have lb keep your DB up to date for either production or development, (or whatever scheme you want).
Irmin (branching + time travel)
Flur.ee (immutable + time travel + graph query)
XTDB (formerly called 'CruxDB') (time travel + query)
TerminusDB (immutable + branching + time travel + Graph Query!)
DoltDB (branching + time-travel + SQL query)
Quadrable (branching + remote state verification)
EdgeDB (no real time travel, but migrations derived by the compiler after schema changes)
Migra (diffing for Postgres schemas/data. Auto-generate migration scripts, auto-sync db state)
ImmuDB (immutable + time-travel)
I've come across this question, as I've got a similar problem, where something approximating a DB based Directory structure, stores 'files', and I need git to manage it. It's distributed, across a cloud, using replication, hence it's access point will be via MySQL.
The gist of the above answers, seem to similarly suggest an alternative solution to the problem asked, which kind of misses the point, of using Git to manage something in a Database, so I'll attempt to answer that question.
Git is a system, which in essence stores a database of deltas (differences), which can be reassembled, in order, to reproduce a context. The normal usage of git assumes that context is a filesystem, and those deltas are diff's in that file system, but really all git is, is a hierarchical database of deltas (hierarchical, because in most cases each delta is a commit with at least 1 parents, arranged in a tree).
As long as you can generate a delta, in theory, git can store it. The problem is normally git expects the context, on which it's generating delta's to be a file system, and similarly, when you checkout a point in the git hierarchy, it expects to generate a filesystem.
If you want to manage change, in a database, you have 2 discrete problems, and I would address them separately (if I were you). The first is schema, the second is data (although in your question, you state data isn't something you're concerned about). A problem I had in the past, was a Dev and Prod database, where Dev could take incremental changes to the schema, and those changes had to be documented in CVS, and propogated to live, along with additions to one of several 'static' tables. We did that by having a 3rd database, called Cruise, which contained only the static data. At any point the schema from Dev and Cruise could be compared, and we had a script to take the diff of those 2 files and produce an SQL file containing ALTER statements, to apply it. Similarly any new data, could be distilled to an SQL file containing INSERT commands. As long as fields and tables are only added, and never deleted, the process could automate generating the SQL statements to apply the delta.
The mechanism by which git generates deltas is diff and the mechanism by which it combines 1 or more deltas with a file, is called merge. If you can come up with a method for diffing and merging from a different context, git should work, but as has been discussed you may prefer a tool that does that for you. My first thought towards solving that is this https://git-scm.com/book/en/v2/Customizing-Git-Git-Configuration#External-Merge-and-Diff-Tools which details how to replace git's internal diff and merge tool. I'll update this answer, as I come up with a better solution to the problem, but in my case I expect to only have to manage data changes, in-so-far-as a DB based filestore may change, so my solution may not be exactly what you need.
There is a great project called Migrations under Doctrine that built just for this purpose.
Its still in alpha state and built for php.
http://docs.doctrine-project.org/projects/doctrine-migrations/en/latest/index.html
Take a look at RedGate SQL Source Control.
http://www.red-gate.com/products/sql-development/sql-source-control/
This tool is a SQL Server Management Studio snap-in which will allow you to place your database under Source Control with Git.
It's a bit pricey at $495 per user, but there is a 28 day free trial available.
NOTE
I am not affiliated with RedGate in any way whatsoever.
I've released a tool for sqlite that does what you're asking for. It uses a custom diff driver leveraging the sqlite projects tool 'sqldiff', UUIDs as primary keys, and leaves off the sqlite rowid. It is still in alpha so feedback is appreciated.
Postgres and mysql are trickier, as the binary data is kept in multiple files and may not even be valid if you were able to snapshot it.
https://github.com/cannadayr/git-sqlite
I want to make something similar, add my database changes to my version control system.
I am going to follow the ideas in this post from Vladimir Khorikov "Database versioning best practices". In summary i will
store both its schema and the reference data in a source control system.
for every modification we will create a separate SQL script with the changes
In case it helps!
You can't do it without atomicity, and you can't get atomicity without either using pg_dump or a snapshotting filesystem.
My postgres instance is on zfs, which I snapshot occasionally. It's approximately instant and consistent.
I think X-Istence is on the right track, but there are a few more improvements you can make to this strategy. First, use:
$pg_dump --schema ...
to dump the tables, sequences, etc and place this file under version control. You'll use this to separate the compatibility changes between your branches.
Next, perform a data dump for the set of tables that contain configuration required for your application to operate (should probably skip user data, etc), like form defaults and other data non-user modifiable data. You can do this selectively by using:
$pg_dump --table=.. <or> --exclude-table=..
This is a good idea because the repo can get really clunky when your database gets to 100Mb+ when doing a full data dump. A better idea is to back up a more minimal set of data that you require to test your app. If your default data is very large though, this may still cause problems though.
If you absolutely need to place full backups in the repo, consider doing it in a branch outside of your source tree. An external backup system with some reference to the matching svn rev is likely best for this though.
Also, I suggest using text format dumps over binary for revision purposes (for the schema at least) since these are easier to diff. You can always compress these to save space prior to checking in.
Finally, have a look at the postgres backup documentation if you haven't already. The way you're commenting on backing up 'the database' rather than a dump makes me wonder if you're thinking of file system based backups (see section 23.2 for caveats).
What you want, in spirit, is perhaps something like Post Facto, which stores versions of a database in a database. Check this presentation.
The project apparently never really went anywhere, so it probably won't help you immediately, but it's an interesting concept. I fear that doing this properly would be very difficult, because even version 1 would have to get all the details right in order to have people trust their work to it.
This question is pretty much answered but I would like to complement X-Istence's and Dana the Sane's answer with a small suggestion.
If you need revision control with some degree of granularity, say daily, you could couple the text dump of both the tables and the schema with a tool like rdiff-backup which does incremental backups. The advantage is that instead of storing snapshots of daily backups, you simply store the differences from the previous day.
With this you have both the advantage of revision control and you don't waste too much space.
In any case, using git directly on big flat files which change very frequently is not a good solution. If your database becomes too big, git will start to have some problems managing the files.
Here is what i am trying to do in my projects:
separate data and schema and default data.
The database configuration is stored in configuration file that is not under version control (.gitignore)
The database defaults (for setting up new Projects) is a simple SQL file under version control.
For the database schema create a database schema dump under the version control.
The most common way is to have update scripts that contains SQL Statements, (ALTER Table.. or UPDATE). You also need to have a place in your database where you save the current version of you schema)
Take a look at other big open source database projects (piwik,or your favorite cms system), they all use updatescripts (1.sql,2.sql,3.sh,4.php.5.sql)
But this a very time intensive job, you have to create, and test the updatescripts and you need to run a common updatescript that compares the version and run all necessary update scripts.
So theoretically (and thats what i am looking for) you could
dumped the the database schema after each change (manually, conjob, git hooks (maybe before commit))
(and only in some very special cases create updatescripts)
After that in your common updatescript (run the normal updatescripts, for the special cases) and then compare the schemas (the dump and current database) and then automatically generate the nessesary ALTER Statements. There some tools that can do this already, but haven't found yet a good one.
What I do in my personal projects is, I store my whole database to dropbox and then point MAMP, WAMP workflow to use it right from there.. That way database is always up-to-date where ever I need to do some developing. But that's just for dev! Live sites is using own server for that off course! :)
Storing each level of database changes under git versioning control is like pushing your entire database with each commit and restoring your entire database with each pull.
If your database is so prone to crucial changes and you cannot afford to loose them, you can just update your pre_commit and post_merge hooks.
I did the same with one of my projects and you can find the directions here.
That's how I do it:
Since your have free choise about DB type use a filebased DB like e.g. firebird.
Create a template DB which has the schema that fits your actual branch and store it in your repository.
When executing your application programmatically create a copy of your template DB, store it somewhere else and just work with that copy.
This way you can put your DB schema under version control without the data. And if you change your schema you just have to change the template DB
We used to run a social website, on a standard LAMP configuration. We had a Live server, Test server, and Development server, as well as the local developers machines. All were managed using GIT.
On each machine, we had the PHP files, but also the MySQL service, and a folder with Images that users would upload. The Live server grew to have some 100K (!) recurrent users, the dump was about 2GB (!), the Image folder was some 50GB (!). By the time that I left, our server was reaching the limit of its CPU, Ram, and most of all, the concurrent net connection limits (We even compiled our own version of network card driver to max out the server 'lol'). We could not (nor should you assume with your website) put 2GB of data and 50GB of images in GIT.
To manage all this under GIT easily, we would ignore the binary folders (the folders containing the Images) by inserting these folder paths into .gitignore. We also had a folder called SQL outside the Apache documentroot path. In that SQL folder, we would put our SQL files from the developers in incremental numberings (001.florianm.sql, 001.johns.sql, 002.florianm.sql, etc). These SQL files were managed by GIT as well. The first sql file would indeed contain a large set of DB schema. We don't add user-data in GIT (eg the records of the users table, or the comments table), but data like configs or topology or other site specific data, was maintained in the sql files (and hence by GIT). Mostly its the developers (who know the code best) that determine what and what is not maintained by GIT with regards to SQL schema and data.
When it got to a release, the administrator logs in onto the dev server, merges the live branch with all developers and needed branches on the dev machine to an update branch, and pushed it to the test server. On the test server, he checks if the updating process for the Live server is still valid, and in quick succession, points all traffic in Apache to a placeholder site, creates a DB dump, points the working directory from 'live' to 'update', executes all new sql files into mysql, and repoints the traffic back to the correct site. When all stakeholders agreed after reviewing the test server, the Administrator did the same thing from Test server to Live server. Afterwards, he merges the live branch on the production server, to the master branch accross all servers, and rebased all live branches. The developers were responsible themselves to rebase their branches, but they generally know what they are doing.
If there were problems on the test server, eg. the merges had too many conflicts, then the code was reverted (pointing the working branch back to 'live') and the sql files were never executed. The moment that the sql files were executed, this was considered as a non-reversible action at the time. If the SQL files were not working properly, then the DB was restored using the Dump (and the developers told off, for providing ill-tested SQL files).
Today, we maintain both a sql-up and sql-down folder, with equivalent filenames, where the developers have to test that both the upgrading sql files, can be equally downgraded. This could ultimately be executed with a bash script, but its a good idea if human eyes kept monitoring the upgrade process.
It's not great, but its manageable. Hope this gives an insight into a real-life, practical, relatively high-availability site. Be it a bit outdated, but still followed.
Update Aug 26, 2019:
Netlify CMS is doing it with GitHub, an example implementation can be found here with all information on how they implemented it netlify-cms-backend-github
I say don't. Data can change at any given time. Instead you should only commit data models in your code, schema and table definitions (create database and create table statements) and sample data for unit tests. This is kinda the way that Laravel does it, committing database migrations and seeds.
I would recommend neXtep (Link removed - Domain was taken over by a NSFW-Website) for version controlling the database it has got a good set of documentation and forums that explains how to install and the errors encountered. I have tested it for postgreSQL 9.1 and 9.3, i was able to get it working for 9.1 but for 9.3 it doesn't seems to work.
Use a tool like iBatis Migrations (manual, short tutorial video) which allows you to version control the changes you make to a database throughout the lifecycle of a project, rather than the database itself.
This allows you to selectively apply individual changes to different environments, keep a changelog of which changes are in which environments, create scripts to apply changes A through N, rollback changes, etc.
I'd like to put the entire database under version control, what
database engine can I use so that I can put the actual database under
version control instead of its dump?
This is not database engine dependent. By Microsoft SQL Server there are lots of version controlling programs. I don't think that problem can be solved with git, you have to use a pgsql specific schema version control system. I don't know whether such a thing exists or not...
Use a version-controlled database, of which there are now several.
https://www.dolthub.com/blog/2021-09-17-database-version-control/
These products don't apply version control on top of another type of database -- they are their own database engines that support version control operations. So you need to migrate to them or start building on them in the first place.
I write one of them, DoltDB, which combines the interfaces of MySQL and Git. Check it out here:
https://github.com/dolthub/dolt
I wish it were simpler. Checking in the schema as a text file is a good start to capture the structure of the DB. For the content, however, I have not found a cleaner, better method for git than CSV files. One per table. The DB can then be edited on multiple branches and merges extremely well.

How to automate source control with Oracle database

I work in an Oracle instance that has hundreds of schemas and multiple developers. We have a development instance where developers can integrate their work before test or production.
We want to have source control for all the DDL run in this integrated development database. Currently this is done through a product Red Gate which we run manually after we make a change to the database. Redgate finds the changes between what is in the schema and what was last checked into source control and makes a script of the differences and puts this into source control.
The problem however is of course that running regdate can take some time and people run it infrequently or not at all for small changes. Also redgate will only look in one schema at a time and it would be VERY time consuming to manually run it against all schemas to guarantee that they are up to date. However if the source controlled code cannot be relied upon it becomes less useful...
What would seem to be ideal would be to have some software that could periodically (even once a day), or when triggered by DDL being run, update the source control (preferably github as this is used by other teams) from all the schemas.
I cannot seem to see any existing software which can be simply used to do this.
Is there a problem with doing this? (there is no need to address multiple developers overwriting each others work on the same day as we have this covered in a separate process) Is anyone doing this? Can anyone recommend a way to do this?
We do this with help of a PL/SQL function, a python script and a shell script:
The PL/SQL function can generate the DDL of a whole schema and returns this as CLOB
The python script connects to the database, fetches the DDL and stores it in files
The shell script runs the Source Control to add the modifications (we use Bazaar here).
You can see the scripts on PasteBin:
The PL/SQL function is here: http://pastebin.com/AG2Fa9zL
The python program (schema_exporter.py): http://pastebin.com/nd8Lf0gK
The shell script:
The shell script:
python schema_exporter.py
d=$(date +%Y-%m-%d__%H_%M_%S)
bzr add
bzr st | grep -q -E 'added|modified' && commit -m "Database objects on $d"
exit 0
This shell script is configured to run from cron every day.
Being in the database version control space for 5 years (as director of product management at DBmaestro) and having worked as a DBA for over two decades, I can tell you the simple fact that you cannot treat the database objects as you treat your Java, C# or other files and save the changes in simple DDL scripts.
There are many reasons and I'll name a few:
Files are stored locally on the developer’s PC and the change s/he
makes do not affect other developers. Likewise, the developer is not
affected by changes made by her colleague. In database this is
(usually) not the case and developers share the same database
environment, so any change that were committed to the database affect
others.
Publishing code changes is done using the Check-In / Submit Changes /
etc. (depending on which source control tool you use). At that point,
the code from the local directory of the developer is inserted into
the source control repository. Developer who wants to get the latest
code need to request it from the source control tool. In database the
change already exists and impacts other data even if it was not
checked-in into the repository.
During the file check-in, the source control tool performs a conflict
check to see if the same file was modified and checked-in by another
developer during the time you modified your local copy. Again there
is no check for this in the database. If you alter a procedure from
your local PC and at the same time I modify the same procedure with
code form my local PC then we override each other’s changes.
The build process of code is done by getting the label / latest
version of the code to an empty directory and then perform a build –
compile. The output are binaries in which we copy & replace the
existing. We don't care what was before. In database we cannot
recreate the database as we need to maintain the data! Also the
deployment executes SQL scripts which were generated in the build
process.
When executing the SQL scripts (with the DDL, DCL, DML (for static
content) commands) you assume the current structure of the
environment match the structure when you create the scripts. If not,
then your scripts can fail as you are trying to add new column which
already exists.
Treating SQL scripts as code and manually generating them will cause
syntax errors, database dependencies errors, scripts that are not
reusable which complicate the task of developing, maintaining,
testing those scripts. In addition, those scripts may run on an
environment which is different from the one you though it would run
on.
Sometimes the script in the version control repository does not match
the structure of the object that was tested and then errors will
happen in production!
There are many more, but I think you got the picture.
What I found that works is the following:
Use an enforced version control system that enforces
check-out/check-in operations on the database objects. This will
make sure the version control repository matches the code that was
checked-in as it reads the metadata of the object in the check-in
operation and not as a separated step done manually. This also allow
several developers to work in parallel on the same database while
preventing them to accidently override each other code.
Use an impact analysis that utilize baselines as part of the
comparison to identify conflicts and identify if a difference (when
comparing the object's structure between the source control
repository and the database) is a real change that origin from
development or a difference that was origin from a different path and
then it should be skipped, such as different branch or an emergency
fix.
Use a solution that knows how to perform Impact Analysis for many
schemas at once, using UI or using API in order to eventually
automate the build & deploy process.
An article I wrote on this was published here, you are welcome to read it.
To me it seems like your way of working is backwards: developers run DDL against the DB in an unordered fashion and then you need an automated tool for inferring the changes (and the DDL) that was run.
The process would be in better control if you did the following instead:
Developers write DDL as SQL scripts, preferably using a migration tool such as Flyway (http://flywaydb.org/documentation/migration/sql.html).
Migration scripts are checked into version control
Migration scripts are periodically run against the DB (e.g. by the migration tool)
In this workflow, the DB would only get altered through automated migration scripts and no-one is allowed to do changes manually. Could this work for you?
(I develop the Oracle tools for Redgate)
Actually using the tools you can already what I think you're asking for using Schema Compare for Oracle.
You can compare multiple schemas either in the UI or via the command line - I think what you're after is automating the command line tool which can create difference scripts, sync between source and destination (live, snapshot or scripts) and generate reports.
You can automate the command line to sync to a scripts folder which is your source code checkout and then subsequently run a command to commit the changes.
I think that's all good :)
We built a commerical tool that bridges Oracle with Git. It helps you manage your database objects with Git. Basically, the database becomes the working directory for the developer. You can perform git operations in the database such as reset, commit, branch, merge etc... and the database code is updated automatically. It might be worth taking a look: https://www.gitora.com

Add version control to existing SQL Server database [closed]

Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 7 years ago.
Improve this question
I am part of a development team currently working with a database that does not have any kind of source control. We work with SQL Server 2008 R2 and have always managed the DB directly with SSMS. It now has ~340 tables and ~1600 stored procedures, plus a few triggers and views, so it is not a small DB.
My goal is to have the DB under version control, so I have been reading articles, like Scott Allen's series and many old SO related questions. But I am still unable to decide on how to proceed.
What I'm thinking of is to script the database schema in one file, then procedures, triggers and views in one file each. Then keep everything versioned under Mercurial. But of course, every member of the team can access SSMS and directly change the schema and procedures, with the possibility that any of us can forget to replicate those changes in the versioned files.
What better options are there? And, did I forget any element worth having source control of? My biggest concern is that most of the literature I found explains how to do version control when creating a new database, but not when it is already old and relatively big.
The General Process
We create a baseline for a particular version (say, v1.0). A baseline includes one complete schema creation script, as well an upgrade script from allowed previous versions, if any (more on that in a moment). So for v1.0, we'd have just one script:
baseline-v1.0.sql
From that baseline, we create incremental change scripts as we work from the previous baseline. These scripts are created in a way that they are reentrant, so that they can be run safely multiple times (where the first time only does any actual work; see the next paragraph on a suggestion how). We just create a file for each change script with the baseline name and a timestamp (which we call the version). So for example, say we create two change scripts after a baseline. We'd have the following files:
baseline-v1.0.sql (for creating new installations)
baseline-v1.0-201211071220.sql (created on Nov. 7, 2012 at 12:20 PM UTC)
baseline-v1.0-201211122019.sql (created on Nov. 12, 2012 at 8:00 PM UTC)
We create a schema_version table that has two columns: baseline and version. baseline is some label (such as v1.0 mentioned above), and version is just a timestamp of when the change script was created (we chose to do this because creating arbitrary version numbers created annoying administrative overhead, where a timestamp was easy to use). So before running the change script, we check to see if the change script has been applied yet, by querying for it by baseline and version. If it's already present, just return out of the script or whatever. Otherwise, apply the change and insert into the schema_version table to mark the change script completed.
Example change script:
-- Created by <developer> on Nov. 7, 2012 at 12:20 PM UTC
declare #schema_baseline varchar(10), #schema_version varchar(12)
set #schema_baseline = 'v1.0'
set #schema_version = '201211071210'
if exists (select 1 from schema_version where baseline = #schema_baseline and version = #schema_version = #schema_version) return 0
-- begin change script
-- place your schema changes here
-- end change script
insert into schema_version(#schema_baseline, #schema_version)
Now, when we actually install the software, we run the relevant baseline script. As we upgrade that version, we just apply the change scripts in order.
When we hit a significant milestone in our product development phase, we create a new baseline. So, we create a new baseline script (again, this is a snapshot of the DB as a baseline), plus an upgrade script from the previous baseline. So let's say we have a new baseline, v2.0, we'd have the following files:
baseline-v2.0.sql (for creating new installations)
baseline-v2.0-upgrade-v1.0.sql (for upgrading from v1.0)
Then the process continues.
How We Apply Changes
The scripts are all kept in source control. We do have a tool that packages these files and automatically upgrades databases, which our support and installation teams use. The tool figures out the current baseline of the target database, and asks the user if they wish to upgrade to the baseline in the package. If they do, and there is a valid upgrade path from the current version, it applies the upgrade script, and updates the schema_version.baseline, and deletes all entries for change scripts from the previous baseline. If the database is new, it applies the regular baseline script. Either way, after the baseline is achieved, it applies all change scripts from the baseline that are present in the package, one at a time, in order, in a transaction. If a particular change script fails, it rolls back the last set of changes and errors out. We look at the log, fix any issues, then rerun the package again. At that point, it should just pick up at the last change script that succeeded, saving time.
Automation and Diff Tools
We do not allow diff tools to upgrade production databases directly. It's just too risky. We do use diff tools, of course, to help create our upgrade and change scripts, but once we have them, we comb through them, massage them, test them, etc., then create the upgrade or change script according to the specs above. We do use tools/shell scripts to create the change script files and put the boiler plate schema_version checking.
Caveats
It's actually pretty straight-forward and it works well. The only time it really gets tricky is with branches. For the most part, branches are handled well. If we need a change script for a particular branch's work, it will fold into the mainline very well once we merge the branch back in. No problem. Where it gets tricky is when two branches try to do similar things, or where one branch relies on another. That's mostly a process and planning issue, though. If we get stuck in such a situation, we just create a new baseline (say v2.1), then update the branches accordingly.
Another thing to keep in mind is if an installation wants to be upgraded from one baseline to another, it has to apply all outstanding changes for the current baseline, before we upgrade to the new one. In other words, we don't let installations jump right from where ever they are to the next baseline (unless, of course, they're already at the most recent version for the current baseline).
I would recommend SQL Server Data Tools and/or a Visual Studio SQL database project. It will reverse engineer your existing DB to code(sql) files that can be version controlled and gives many other niceties (publishing, comparison, etc)
We developed SQL Source Control specifically to solve the problem you describe. It extends SSMS to provide a link between your SQL Server schema objects (and static data) and your existing source control system.
http://www.red-gate.com/products/sql-development/sql-source-control/
If you need any more information, we'd be very pleased to help (contact support#red-gate.com)
There have been many discussions regarding this topic on many developer forums.
What I have done and found to be the simplest and cleanest way to do is this:
Extract every DB object's DDL into its own file, indexes and PKs can go in the same file as the table they belong to. FKs, procedures, views, triggers, anything that can go across multiple tables go in their own file.
Organize the DDL files in dirs per object type (e.g. table, procedure, trigger, view etc.)
For tables holding static reference data (e.g. zip code or state), have a separate file with a bunch of insert statements
Check this directory structure into whatever version control you are using
Write a script that will traverse this directory structure that images your DB, diff it against the actual DB you point to (extracting the schema from the system tables) and apply the diffs using ALTER TABLE statements
In case you have data transformations between releases, e.g. in v1 you had a field FirstAndLastName and in v2 you decided to split it into FirstName and LastName, you will have some bulk data migration/processing statement.
I have successfully managed DB changes in several jobs using several different RDBMSs. I usually use Perl for the script that diffs the DB schema and the DDL files in your image. There are some assumptions to this method and one of them is that you never make any changes to the DB directly in the DB but in your DDL files and then run the script to apply it. If you do it the other way, they will be undone when you run the script. So it requires some team agreement and discipline. Your milage may vary.
Now, if there is a FOSS tool out there that will do this for you, by all means use that rather than devising your own. I've been doing things this way for more than 10 yrs
Our Sql Historian source control system can help folks with this problem, especially in the situation you mention where teammates"forget" to check in code after they've updated the server.
It sits in the background and records all changes made to your db objects into source control, without users needing to check anything in. Think of it like an airplane blackbox recorder, staying out of the way until you need it.

sql server batch database alters, batch database changes - best and safest way

We have a small development team of 5 developers working on a large enterprise level web based asp.net/c# system.
We do a lot of database updates which include stored procedure creations and alters as well as new table creation, column creation, record inserts, record updates and so on and so forth.
Today all of the developers place all change scripts in one large sql change script file that gets ran on our Test and Production environments. So this single file contains stored proc alters and record inserts, updates etc etc. The file can end up being quite lengthy as we may only do a test or production release every 1 to 2 months.
The problem that I am currently facing is this:
Once in a while there is a script error that may occur at any given location in this large "batch change script". Perhaps an insert fails or perhaps an alter fails for a proc for instance.
When this occurs, it is very difficult to tell what changes succeeded and what failed on the database.
Sometimes even if one alter fails for instance code will continue to execute throughout the script and sometimes it will stop execution and nothing further gets ran.
So I end up manually checking procs and records today to see what actually worked and what actually did not and this is a bit painstaking.
I was hoping I could roll up this entire change script into one big transaction so that if any problem occurred I could just roll every change back, but that does not appear to be possible with batch scripts like this in sql server.
So, then I tried to backup the databases before I ran the scripts so that if an error occurred I could simply restore the db, fix the problem and then re-run the fixed script. However in order to restore a database I have to turn off our database mirroring so this is also not totally ideal.
So my question is, what is the safest way to run batch scripts on a production database?
Is there some way that I can wrap the entire script in a transaction that i can roll back that I am not seeing?
Would it possibly be better for us to track and run separate script files so that if 1 file fails we can just shove it off in a failed directory to be looked at and continue running all other files?
Looking for advice and expertise.
thank you for your time.
Matt
The batch script should be run on your QC database first so that any errors are picked up before production.
The QC database should be identical to production or as close as it can be to identical.
Each script should be trapping for an error and reporting the name of the script along with the location of the error using print statements, then if an error occurs when applying to production you at least have the name of the script and the location of the error within the script.
If your QC database is identical or very close, productions errors should be very rare.

Do you put your database static data into source-control ? How?

I'm using SQL-Server 2008 with Visual Studio Database Edition.
With this setup, keeping your schema in sync is very easy. Basically, there's a 'compare schema' tool that allow me to sync the schema of two databases and/or a database schema with a source-controlled creation script folder.
However, the situation is less clear when it comes to data, which can be of three different kind :
static data referenced in the code. typical example : my users can change their setting, and their configuration is stored on the server. However, there's a system-wide default value for each setting that is used in case the user didn't override it. The table containing those default settings grows as more options are added to the program. This means that when a new feature/option is checked in, the system-wide default setting is usually created in the database as well.
static data. eg. a product list populating a dropdown list. The program doesn't rely on the existence of a specific product in the list to work. This can be for example a list of unicode-encoded products that should be deployed in production when the new "unicode version" of the program is deployed.
other data, ie everything else (logs, user accounts, user data, etc.)
It seems obvious to me that my third item shouldn't be source-controlled (of course, it should be backuped on a regular basis)
But regarding the static data, I'm wondering what to do.
Should I append the insert scripts to the creation scripts? or maybe use separate scripts?
How do I (as a developer) warn the people doing the deployment that they should execute an insert statement ?
Should I differentiate my two kind of data? (the first one being usually created by a dev, while the second one is usually created by a non-dev)
How do you manage your DB static data ?
I have explained the technique I used in my blog Version Control and Your Database. I use database metadata (in this case SQL Server extended properties) to store the deployed application version. I only have scripts that upgrade from version to version. At startup the application reads the deployed version from the database metadata (lack of metadata is interpreted as version 0, ie. nothing is yet deployed). For each version there is an application function that upgrades to the next version. Usually this function runs an internal resource T-SQL script that does the upgrade, but it can be something else, like deploying a CLR assembly in the database.
There is no script to deploy the 'current' database schema. New installments iterate trough all intermediate versions, from version 1 to current version.
There are several advantages I enjoy by this technique:
Is easy for me to test a new version. I have a backup of the previous version, I apply the upgrade script, then I can revert to the previous version, change the script, try again, until I'm happy with the result.
My application can be deployed on top of any previous version. Various clients have various deployed version. When they upgrade, my application supports upgrade from any previous version.
There is no difference between a fresh install and an upgrade, it runs the same code, so I have fewer code paths to maintain and test.
There is no difference between DML and DDL changes (your original question). they all treated the same way, as script run to change from one version to next. When I need to make a change like you describe (change a default), I actually increase the schema version even if no other DDL change occurs. So at version 5.1 the default was 'foo', in 5.2 the default is 'bar' and that is the only difference between the two versions, and the 'upgrade' step is simply an UPDATE statement (followed of course by the version metadata change, ie. sp_updateextendedproperty).
All changes are in source control, part of the application sources (T-SQL scripts mostly).
I can easily get to any previous schema version, eg. to repro a customer complaint, simply by running the upgrade sequence and stopping at the version I'm interested in.
This approach saved my skin a number of times and I'm a true believer now. There is only one disadvantage: there is no obvious place to look in source to find 'what is the current form of procedure foo?'. Because the latest version of foo might have been upgraded 2 or 3 versions ago and it wasn't changed since, I need to look at the upgrade script for that version. I usually resort to just looking into the database and see what's in there, rather than searching through the upgrade scripts.
One final note: this is actually not my invention. This is modeled exactly after how SQL Server itself upgrades the database metadata (mssqlsystemresource).
If you are changing the static data (adding a new item to the table that is used to generate a drop-down list) then the insert should be in source control and deployed with the rest of the code. This is especially true if the insert is needed for the rest of the code to work. Otherwise, this step may be forgotten when the code is deployed and not so nice things happen.
If static data comes from another source (such as an import of the current airport codes in the US), then you may simply need to run an already documented import process. The import process itself should be in source control (we do this with all our SSIS packages), but the data need not be.
Here at Red Gate we recently added a feature to SQL Data Compare allowing static data to be stored as DML (one .sql file for each table) alongside the schema DDL that is currently supported by SQL Compare.
To understand how this works, here is a diagram that explains how it works.
The idea is that when you want to push changes to your target server, you do a comparison using the scripts as the source data source, which generates the necessary DML synchronization script to update the target. This means you don't have to assume that the target is being recreated from scratch each time. In time we hope to support static data in our upcoming SQL Source Control tool.
David Atkinson, Product Manager, Red Gate Software
I have come across this when developing CMS systems.
I went with appending the static data (the stuff referenced in the code) to the database creation scripts, then a separate script to add in any 'initialisation data' (like countries, initial product population etc).
For the first two steps, you could consider using an intermediate format (ie XML) for the data, then using a home grown tool, or something like CodeSmith to generate the SQL, and possible source files as well, if (for example) you have lookup tables which relate to enumerations used in the code - this helps enforce consistency.
This has another benefit that if the schema changes, in many cases you don't have to regenerate all your INSERT statements - you just change the tool.
I really like your distinction of the three types of data.
I agree for the third.
In our application, we try to avoid putting in the database the first, because it is duplicated (as it has to be in the code, the database is a duplicate). A secondary benefice is that we need no join or query to get access to that value from the code, so this speed things up.
If there is additional information that we would like to have in the database, for example if it can be changed per customer site, we separate the two. Other tables can still reference that data (either by index ex: 0, 1, 2, 3 or by code ex: EMPTY, SIMPLE, DOUBLE, ALL).
For the second, the scripts should be in source-control. We separate them from the structure (I think they typically are replaced as time goes, while the structures keeps adding deltas).
How do I (as a developer) warn the people doing the deployment that they should execute an insert statement ?
We have a complete procedure for that, and a readme coming with each release, with scripts and so on...
First off, I have never used Visual Studio Database Edition. You are blessed (or cursed) with whatever tools this utility gives you. Hopefully that includes a lot of flexibility.
I don't know that I'd make that big a difference between your type 1 and type 2 static data. Both are sets of data that are defined once and then never updated, barring subsequent releases and updates, right? In which case the main difference is in how or why the data is as it is, and not so much in how it is stored or initialized. (Unless the data is environment-specific, as in "A" for development, "B" for Production. This would be "type 4" data, and I shall cheerfully ignore it in this post, because I've solved it useing SQLCMD variables and they give me a headache.)
First, I would make a script to create all the tables in the database--preferably only one script, otherwise you can have a LOT of scripts lying about (and find-and-replace when renaming columns becomes very awkward). Then, I would make a script to populate the static data in these tables. This script could be appended to the end of the table script, or made it's own script, or even made one script per table, a good idea if you have hundreds or thousands of rows to load. (Some folks make a csv file and then issue a BULK INSERT on it, but I'd avoid that is it just gives you two files and a complex process [configuring drive mappings on deployment] to manage.)
The key thing to remember is that data (as stored in databases) can and will change over time. Rarely (if ever!) will you have the luxury of deleting your Production database and replacing it with a fresh, shiny, new one devoid of all that crufty data from the past umpteen years. Databases are all about changes over time, and that's where scripts come into their own. You start with the scripts to create the database, and then over time you add scripts that modify the database as changes come along -- and this applies to your static data (of any type) as well.
(Ultimately, my methodology is analogous to accounting: you have accounts, and as changes come in you adjust the accounts with journal entries. If you find you made a mistake, you never go back and modify your entries, you just make a subsequent entries to reverse and fix them. It's only an analogy, but the logic is sound.)
The solution I use is to have create and change scripts in source control, coupled with version information stored in the database.
Then, I have an install wizard that can detect whether it needs to create or update the db - the update process is managed by picking appropriate scripts based on the stored version information in the database.
See this thread's answer. Static data from your first two points should be in source control, IMHO.
Edit: *new
all-in-one or a separate script? it does not really matter as long as you (dev team) agree with your deployment team. I prefer to separate files, but I still can always create all-in-one.sql from those in the proper order [Logins, Roles, Users; Tables; Views; Stored Procedures; UDFs; Static Data; (Audit Tables, Audit Triggers)]
how do you make sure they execute it: well, make it another step in your application/database deployment documentation. If you roll out application which really needs specific (new) static data in the database, then you might want to perform a DB version check in your application. and you update the DB_VERSION to your new release number as part of that script. Then your application on a start-up should check it and report an error if the new DB version is required.
dev and non-dev static data: I have never seen this case actually. More often there is real static data, which you might call "dev", which is major configuration, ISO static data etc. The other type is default lookup data, which is there for users to start with, but they might add more. The mechanism to INSERT these data might be different, because you need to ensure you do not destoy (power-)user-created data.