Related
This might be too broad, but it's a problem I'm having a bear of a time dealing with. We have an application that we distribute to our end users. It's running on top of a derby back end. We can push out code changes fairly easily, it'll go out to our server, see there's a new version, download, overwrite old code, and reboot.
But, as we change our code, we also alter the schema of the derby database. We don't have great methods to update this. Currently we can push SQL updates via FTP. When the program is connected to the internet, it looks for new SQL files, downloads them and runs.
Unfortunately a lot of our clients have limited Internet access, so they get these updates intermittently. Sometimes because they changes are big enough, their local DB schema gets out of sync with what we want. Or they get the code changes via CD but not the SQL changes (someone mails them the CD).
What I've been trying to do is create a SOAP service that can serve up XML representations of the schema. It's been a huge PITA to develop so far.
What are some methods people are currently using to maintain databases like this? I feel like I'm not the first to do this, so there might be better ways than what I'm doing.
Based on some comments here, here's an update:
Basically, I think we screwed ourselves early on by not adhering to a strict versioning of the DB, so I don't know how everyone's DB is at. A lot of people got custom installs built (groan at will). I need a tool that can tell the differences between their DB and a "official" copy.
I have a tool built, it kind of works, but there's so…many…things to keep track of.
Can you distribute the DB changes as part of the code changes? Then, when the app restarts, it checks if it needs to run any updates on the DB.
Obviously, you'll need to version the DB schema to avoid applying the same update more than once.
I know some applications that do this (mostly in Ruby, but also in Java).
If you already have an update mechanism in place in your application that can download a program to alter the installed source code, why not package and run the schema changes as a part of that upgrade process? I would just run the updates as a part of the Java application then.
My team at work handles these changes by using the MyBatis Migration tool, which represents each schema change as a single migration script which contains the "make change" and "rollback" steps. A changelog table is stored in the database which lists which updates have been applied to that database, which makes it easy for the migrate command to determine which updates it needs to apply when run. This specific tool is probably only really useful when you control the database and have the ability to run shell commands and scripts to alter the database, but you can use the same concepts in your approach - package each schema change as an atomic unit and run them from within your program to bring the schema up to the current version, which you can track in the db itself.
You'll need a table containing the version of the database that the user is running, and then you'll need code to upgrade from version n to version n+1. Assuming you have a database user that has access to do schema changes, you can apply schema changes the same way you're now applying code changes.
I'm writing an application that is using a database (currently MySQL 4) to store data.
It is likely that I will make changes to this in the form of updates later to add additional data. Updating the application is simple, it essentially comes down to overwriting the program files with the new ones. However how do I go about updating the database schema?
The database is remote and so my application might exist in several places, so simply dumping the ALTER and CREATE statements in an installer would result in the changes being made multiple times, and I have been asked explicitly for an automatic solution that allows for the application copies to be updated over a transition period, and for schema updates to be automatic.
I considered examining the schema at start-up to look for missing tables and columns, and adding them as needed, however this does not seem like a clean solution. I also considered putting some kind of “schema version” number on the database, but can’t see any way to do this short of a single row table with an int “Version” column which doesn’t seem a good way either.
I can highly recommend Liquibase. It really does work - I've used it and was very impressed.
Essentially, it keeps its own log of statements run on a database and runs them only if not already run/needed. It is XML driven and allows you to use optional pre- and post-execution statements and conditions. You check your XML files into your source control and invoke it from your build tool. It's even suitable for driving production releases.
It's magic.
Rather than rolling your own system for versioning your database it's probably worth looking into an existing framework that will manage it for you.
I use liquibase and have integrated into my build using the maven plugin. Worth checking out!
Just as you proposed, add a table where you store the current version of the database schema. Then you only have to apply the changes between your last schema update and the new release, and set the new version number accordingly. I've done this to update our production database about 300 times, it just works.
I was wondering what are the best practices in order to write SQL scripts to set up databases for production and/or development, for instance:
Should I include the CREATE DATABASE statement?
Should I create users for the database in the same script?
Is correct to disable FK check before executing the body of the script?
May I include the hole script in a transaction?
Is better to generate 1 script per database than one script for all of them?
Thanks!
The problem with your question is is hard to answer as it depends on the way the scripts are used in what you are trying to achieve. you also don't say which DB server you are using as there are tools provided which can make some tasks easier.
Taking your points in order, here are some suggestions, which will probably be very different to everyone elses :)
Should I include the CREATE DATABASE
statement?
What alternative are you thinking of using? If your question is should you put the CREATE DATABASE statement in the same script as the table creation it depends. When developing DB I use a separate create DB script as I have a script to drop all objects and so I don't need to create the database again.
Should I create users for the database in the same script?
I wouldn't, simply because the users may well change but your schema has not. Might as well manage those changes in a smaller script.
Is correct to disable FK check before executing the body of the script?
If you are importing the data in an attempt to recover the database then you may well have to if you are using auto increment IDs and want to keep the same values. Also you may end up importing the tables "out of order" an not want checks performed.
May I include the whole script in a transaction?
Yes, you can, but again it depends on the type of script you are running. If you are importing data after rebuilding a db then the whole import should work or fail. However, your transaction file is going to be huge during the import.
Is better to generate 1 script per database than one script for all of them?
Again, for maintenance purposes it's probably better to keep them separate.
This probably depends what kind of database and how it is used and deployed. I am developing a n-tier standard application that is deployed at many different customer sites.
I do not add a CREATE DATABASE statement in the script. Creating the the database is a part of the installation script which allows the user to choose server, database name and collation
I have no knowledge about the users at my customers sites so I don't add create users statements also the only user that needs access to the database is the user executing the middle tire application.
I do not disable FK checks. I need them to protect the consistency of the database, even if it is I who wrote the body scripts. I use FK to capture my errors.
I do not include the entire script in one transaction. I require from the users to take a backup of the db before they run any db upgrade scripts. For creating of a new database there is nothing to protect so running in a transaction is unnecessary. For upgrades there are sometimes extensive changes to the db. A couple of years ago we switched from varchar to nvarchar in about 250 tables. Not something you would like to do in one transaction.
I would recommend you to generate one script per database and version control the scripts separately.
Direct answers, please ask if you need to expand on any point
* Should I include the CREATE DATABASE statement?
Normally I would include it since you are creating and owning the database.
* Should I create users for the database in the same script?
This is also a good idea, especially if your application uses specific users.
* Is correct to disable FK check before executing the body of the script?
If the script includes data population, then it helps to disable it so that the order is not too important, otherwise you can get into complex scripts to insert (without fk link), create fk record, update fk column.
* May I include the hole script in a transaction?
This is normally not a good idea. Especially if data population is included as the transaction can become quite unwieldy large. Since you are creating the database, just drop it and start again if something goes awry.
* Is better to generate 1 script per database than one script for all of them?
One per database is my recommendation so that they are isolated and easier to troubleshoot if the need arises.
For development purposes it's a good idea to create one script per database object (one script for each table, stored procedure, etc). If you check them into your source control system that way then developers can check out individual objects and you can easily keep track of versions and know what changed and when.
When you deploy you may want to combine the changes for each release into one single script. Tools like Red Gate SQL compare or Visual Studio Team System will help you do that.
Should I include the CREATE DATABASE statement?
Should I create users for the database in the same script?
That depends on your DBMS and your customer.
In an Oracle environment you will probably never be allowed to do such a thing (mainly because in the Oracle world a "database" is something completely different than e.g. in the PostgreSQL or MySQL world).
Sometimes the customer will have a DBA that won't let you create databases (or schemas or users - depending on the DBMS in use). So you will need to supply that information to the DBA in order for him/her to prepare the environment for your script.
May I include the hole script in a transaction?
That totally depends on the DBMS that you are using.
Some DBMS don't support transactional DDL and will implicitely commit any transaction when you execute a DDL statement, so you need to consider the order of your installation script.
For populating the tables with data I would definitely try to do that in a single transaction, but again this depends on your DBMS.
Some DBMS are faster if you commit only once or very seldomly (Oracle and PostgreSQL fall into this category) but will slow down if you commit more often.
Other DBMS handle smaller but more transactions better and will slow down if the transactions get too big (SQL Server and MySQL tend to fall into that direction)
The best practices will differ considerably on whether it is the first time set-up or a new version being pushed. For the first time set-up yes you need create database and create table scripts. For a new version, you need to script only the changes from the previous version, so no create database and no create table unless it is a new table. Now you need alter table statements becasue you don't want to lose the existing data. I do usually write stored procs, functions and views with a drop and create statment as dropping those pbjects doesn't generally affect the underlying data.
I find it best to create all database changes with scripts that are stored in source control under the version. So if a client is new, you run the create version 1.0 scripts, then apply all the other versions in order. If a client is just upgrading from version 1.2 to version 1.3, then you run just the scripts in version 1.3 source control repository. This would also include scripts to populate or add records to lookup tables.
For transactions you may want to break them up into several chunks not to leave a prod database locked in one transaction.
We also write reversal scripts to return to the old version if need be. This makes life easier if you have a part of a change that causes unanticipated problems on prod (usually performance issues).
The Background
My group has 4 SQL Server Databases:
Production
UAT
Test
Dev
I work in the Dev environment. When the time comes to promote the objects I've been working on (tables, views, functions, stored procs) I make a request of my manager, who promotes to Test. After testing, she submits a request to an Admin who promotes to UAT. After successful user testing, the same Admin promotes to Production.
The Problem
The entire process is awkward for a few reasons.
Each person must manually track their changes. If I update, add, remove any objects I need to track them so that my promotion request contains everything I've done. In theory, if I miss something testing or UAT should catch it, but this isn't certain and it's a waste of the tester's time, anyway.
Lots of changes I make are iterative and done in a GUI, which means there's no record of what changes I made, only the end result (at least as far as I know).
We're in the fairly early stages of building out a data mart, so the majority of the changes made, at least count-wise, are minor things: changing the data type for a column, altering the names of tables as we crystallize what they'll be used for, tweaking functions and stored procs, etc.
The Question
People have been doing this kind of work for decades, so I imagine there have got to be a much better way to manage the process. What I would love is if I could run a diff between two databases to see how the structure was different, use that diff to generate a change script, use that change script as my promotion request. Is this possible? If not, are there any other ways to organize this process?
For the record, we're a 100% Microsoft shop, just now updating everything to SQL Server 2008, so any tools available in that package would be fair game.
I should clarify I'm not necessarily looking for diff tools. If that's the best way to sync our environments then it's fine, but if there's a better way I'm looking for that.
An example doing what I want really well are migrations in Ruby on Rails. Dead simple syntax, all changes are well documented automatically and by default, determining what migrations need to run is almost trivially easy. I'd love if there was something similar to this for SQL Server.
My ideal solution is 1) easy and 2) hard to mess up. Rails Migrations are both; everything I've done so far on SQL Server is neither.
Within our team, we handle database changes like this:
We (re-)generate a script which creates the complete database and check it into version control together with the other changes. We have 4 files: tables, user defined functions and views, stored procedures, and permissions. This is completely automated - only a double-click is needed to generate the script.
If a developer has to make changes to the database, she does so on her local db.
For every change, we create update scripts. Those are easy to create: The developer regenerates the db script of his local db. All the changes are now easy to identify thanks to version control. Most changes (new tables, new views etc) can simply be copied to the update script, other changes (adding columns for example) need to be created manually.
The update script is tested either on our common dev database, or by rolling back the local db to the last backup - which was created before starting to change the database. If it passes, it's time to commit the changes.
The update scripts follow a naming convention so everybody knows in which order to execute them.
This works fairly well for us, but still needs some coordination if several developers modify heavily the same tables and views. This doesn't happen often though.
The important points are:
database structure is only modified by scripts, except for the local developer's db. This is important.
SQL scripts are versioned by source control - the db can be created as it was at any point in the past
database backups are created regularly - at least before making changes to the db
changes to the db can be done quickly - because the scripts for those changes are created relatively easily.
However, if you have a lot of long lasting development branches for your projects, this may not work well.
It is by far not a perfect solution, and some special precautions are to be taken. For example, if there are updates which may fail depending on the data present in a database, the update should be tested on a copy of the production database.
In contrast to rails migrations, we do not create scripts to reverse the changes of an update. But this isn't always possible anyway, at least in respect to the data (the content of a dropped column is lost even if you recreate the column).
Version Control and your Database
The root of all things evil is making changes in the UI. SSMS is a DBA tool, not a developer one. Developers must use scripts to do any sort of changes to the database model/schema. Versioning your metadata and having upgrade script from every version N to version N+1 is the only way that is proven to work reliably. It is the solution SQL Server itself deploys to keep track of metadata changes (resource db changes).
Comparison tools like SQL Compare or vsdbcmd and .dbschema files from VS Database projects are just last resorts for shops that fail to do a proper versioned approach. They work in simple scenarios, but I see them all fail spectacularly in serious deployments. One just does not trust a tool to do a change to +5TB table if the tools tries to copy the data...
RedGate sells SQL Compare, an excellent tool to generate change scripts.
Visual Studio also has editions which support database compares. This was formerly called Database Edition.
Where I work, we abolished the Dev/Test/UAT/Prod separation long ago in favor of a very quick release cycle. If we put something broken in production, we will fix it quickly. Our customers are certainly happier, but in the risk avert corporate enterprise, it can be a hard sell.
There are several tools available for you. One is from Red-Gate called SQL Compare. Awesome and highly recommended. SQL Compare will let you do a diff in schemas between two databases and even build the sql change scripts for you.
Note they have been working on a SQL Server source control product for awhile now as well.
Another (if you're a visual studio shop) is the schema and data compare features that is part of Visual Studio (not sure which versions).
Agree that SQL Compare is an amazing tool.
However, we do not make any changes to the database structure or objects that are not scripted and saved in source control just like all other code. Then you know exactly what belongs in the version you are promoting because you have the scripts for that particular version.
It is a bad idea anyway to make structural changes through the GUI. If you havea lot of data, it is far slower than using alter table at least in SQL Server. You only want to use tested scripts to make changes to prod as well.
I agree with the comments made by marapet, where each change must be scripted.
The problem that you may be experiencing, however, is creating, testing and tracking these scripts.
Have a look at the patching engine used in DBSourceTools.
http://dbsourcetools.codeplex.com
It's been specifically designed to help developers get SQL server databases under source-code control.
This tool will allow you to baseline your database at a specific point, and create a named version (v1).
Then, create a deployment target - and increment the named version to v2.
Add patch scripts to the Patches directory for any changes to schema or data.
Finally, check the database and all patches into source-code control, to distribute with devs.
What this gives you is a repeatable process to test all patches to be applied from v1 to v2.
DBSourceTools also has functionality to help you create these scripts, i.e. schema compare or script data tools.
Once you are done, simply send all of the files in the patches directory to your DBA to upgrade from v1 to v2.
Have fun.
Another "Diff" tool for databases:
http://www.xsqlsoftware.com/Product/Sql_Data_Compare.aspx
Keep database version in a versioning table
Keep script file name that was successfully applied
Keep md5 sum of each sql script that has been applied. It should ignore spaces when calculate md5 sum. Must be effective.
Keep info about who applied a script Keep info about when a script was applied
Database should be verified on application start-up
New sql script should be applied automatically
If md5 sum of a script that was already applied is changed, error should be thrown (in a production mode)
When script have been released it must not be changed. It must be
immutable in a production environment.
Script should be written in a way, so it could be applied to different types of database (see liquibase)
Since most ddl statements are auto-committing on most databases, it is best to have a single ddl statement per SQL script.
DDL sql statement should be run in a way, so it can be executed several times without errors. Really helps in a dev mode, when you may edit script several times. For instance, create a new table, only if it does not exist, or even drop table before creating a new one. It will help you in a dev mode, with a script that has not been released, change it, clear md5 sum for this script, rerun it again.
Each sql script should be run in its own transaction.
Triggers/procedures should be dropped and created after each db
update.
Sql script is kept in a versioning system like svn
Name of a script contains date when it was committed, existing (jira) issue id, small description
Avoid adding rollback functionality in scripts (liquibase allow to do that). It makes them more complicated to write and support. If you use exactly one ddl statement per script, and dml statements are run within a
transaction, even failing a script will not be a big trouble to
resolve it
This is the workflow we have been using succesfully:
Development instance: SQL objects are created/updated/deleted in DB using MSSQL Studio and all operations are saved to scritps we include in each version of our code.
Moving to production: We compare schema between dev and prod db using SQL Schema Compare in Microsoft Visual Studio. We update prod using the same tool.
here's a more general question on how you handle database schema changes in a development team.
We are a team of developers and the databases used during development are running locally on everyone's box as we want to avoid the requirement to have web access all the time. So running a single central database instance somewhere is not a real option.
Whenever one of us decides that it is time to extend/change the db schema, we mail database files (MYI/MYD) or SQL files to execute around, or give others instructions on the phone what they need to do to get the changed code running on their local DBs. That's not the perfect approach for sure. The same problem arises when we need to adjust the DB schema on staging or production once a new release is ready.
I was wondering ... how do you guys handle this kind of stuff? For source code, we use SVN.
Really appreciate your input!
Thanks,
Michael
One approach we've used in the past is to script the entire DDL for the database, along with any test/setup data needed. Store that in SVN, then when there's a change, any developer can pull down the changes, drop the database, and rebuild it from the script files.
At the very least you should have the scripts of all the objects in the database (tables, stored procedures, etc) under source control.
I don't think mailing schema changes is a real option for a professional development team.
We had a system on one of my previous teams that was the best I've encountered for dealing with this situation.
The nightly build of the application included a build of a database (SQL Server). The database got built to the Test DB server. Each developer then had a DTS package (this was a while ago, and I'm sure they upgraded to SSIS packages) to pull down that nightly DB build to their local DB environment.
This kept the master copy in one location and put the onus on the developers to keep their local dev databases fresh.
At my work, we deal with pretty large databases that are time-consuming to generate, so for us, starting from scratch with a new DB isn't ideal. Like Harper, we have our DDL in SVN. Additionally, we store a version number in a database table. Every check-in that changes the DB must be accompanied by a script that:
Will upgrade the database schema and modify any existing data appropriately, and
Will update the version number in the database.
Further, we number the scripts and database versions such that a script we've written knows how to upgrade further along a branch or from an older branch to a newer one without any input from the developer (apart from the database name and the directory to the upgrade scripts).
Thus, if I've got a copy of a customer's 4GB DB that's from a year old version and I want to test how their data will work with the version we cut yesterday, I can just run our script and let it handle the upgrades rather than having to start from scratch and redo every INSERT, UPDATE and DELETE performed since the database was created.
We have a non-SQL description of the database schema. When the application starts, it compares the desired database schema with the actual database schema, and performs whatever ADD TABLE, ADD COLUMN, ADD INDEX, etc. statements it needs to do to get the database to look right.
This doesn't handle every case; sometimes you have to delete the database and recreate if if you've changed something that the schema resolver can't handle, but most of the time we don't need to worry about it.
I'd certainly keep the database schema in source code control.
At my present job, every time there's a schema change, we write the SQL for the change (alter table xyz add column ...) and put it in SVN. Then developers can update test databases by running this script. It's pretty clumsy but it works.
At a previous job I wrote some code that at application start-up would automatically compare the actual database schema to what it expected, and if it was not up to date perform the updates. Mostly this was done for deployment reasons: When we shipped new copies of the software, it would then automatically update the user's database. But it was also handy for developers.
I think there should be some generic SQL tool to do this. Maybe there is, but I've never seen one.