I am a computer science student and yesterday in my class on database management systems and talking about distributed systems and good practices he said that we can reference other instances through something he called 'connetion stream' or 'connetion String 'I really did not quite understand what he said. But I raised the following question:

I theoretically own a license db, application X and application Y, are totally different applications and only share the access user. In the first database are stored company information, license and access users; In database Y and X are different data that need to share the same user that is in the license database.

I tried in SQL Server as follows:

USE Licenses;
 Id int identity(1, 1),
 Name varchar(100) unique not null,
 //Other user information
 primary key(Id)

CREATE TABLE XInformation(
 Id int identity(1, 1),
 //Other X information
 PersonId int not null references Licenses..Persons(Id)

CREATE TABLE YInformation(
 Id int identity(1, 1),
 //Other Y information
 PersonId int not null references Licenses..Persons(Id)

But I get an error saying that sql server does not support foreign keys between databases.

But if I create these tables without referencing I can query them like this:

SELECT * from Persons as p inner join X..XInformation as xi on p.Id = xi.PersonId;

What is the correct way to do this referencing?

How would a dba solve this case?

Can I only reference other databases in Views?

Thanks in advance.


Unfortunate, there isn't a way to directly create the FK constraint you'd like. Run through some alternatives instead.

(1) Is it really necessary to have those two tables in different databases? Is there a reason why you can't just move the table? You won't be able to enforce any rules on schema changes unless you have the tables in the same db. Push and prod developers to redo it 'the right way'.

(2) Workaround: Create a trigger on insert/update to the referencing table, which does the test FK constraint usually does. Create a trigger on update/delete on the referenced table to implement cascading if needed. You can create an additional DDL trigger that prevents changes to the referenced table's structure. The problem though is that nothing fully prevents table structure changes before the FK is dropped. Another problem can occur when the two databases are not restored together - leading to possible inconsistent configurations.

(3) Create synchronization/replication scheme so that a reasonably up to date copy of the referenced table can be kept in the database where referencing table is located. Can either be a job that copies new rows (Change tracking in mssql) or use SSIS to build an ETL package that incrementally loads new rows into the copy on the referenced table. If the two databases are on separate servers, there are more options using replication features of MSSQL. Loading data into a database local copy is another chance to check schema, but again you can't prevent changes on the referenced table while the FK exists.

For performance, it should still be using the indexes.

Cross database querying are not uncommon, nor are cross server distributed queries. DBA's best friend are scripts and automation, and you can script checks to make sure there's nothing amiss that would break your cross db/server queries and shoot you an email otherwise. There are also other ways to set up alerts (on ddl triger, etc.,) so that you know if there's going to be a problem.

As for your question regarding views, I'm not sure what you mean. You can access tables on another db from within a view, but the constraint part of the FK relationship still can't be directly maintained on a table from another db. You can't schemabind a view using tables from different database either.

  • Sorry, but as said on the database study yet, could you explain what exactly would be SSIS and ETL? The other terms are already more accustomed. I searched the internet but the idea remains unclear about this – Márcio Eric Aug 3 '17 at 17:19
  • Ahh, yes, sorry. SQL Server Integration Services is a Visual Studio based development environment for building Exchange, Transfer and Load operations. It's a convenient IDE to implement code to move data between different data sources/databases and apply any transformations (rename fields, aggregating, etc.,) needed during that process. – Xingzhou Liu Aug 3 '17 at 17:29
  • docs.microsoft.com/en-us/sql/integration-services/…. If you're using express edition and you may need to upgrade to standard. Trial lasts 6 months and lets you reinstall. – Xingzhou Liu Aug 3 '17 at 17:33
  • 1
    Or Dev, which is free and lasts forever ;) – Kevin3NF Aug 3 '17 at 18:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.