Question

Two applications are using the same MSSQL database.

My application is using sqlalchemy with pymssql and wants to receive a notification when a row is added to a specific table by another application.

What is the correct way of doing this?

Answer 1

Lets call your application “consumer”, and the other application “producer”.

There are several approaches, the best fit will depend on how many records, producers and consumers you are expecting.

1. for small numbers of records, producer and consumer instances, you can add an “ack” field to the table with false as the default value. New entries should have this field set to false, so other application can set “ack” to true after processing new entries.
2. you can have a table acting as a queue, where producer instances insert new records, and consumer instances delete the records after copying to the final destination table.
3. for maximum scalability you can use something like AMQP to manage the communication between consumers and producers

Be aware that any algorithm that involves polling the database for changes has potential to tax heavily on the underlying DBMS.

[update]

Suggestion from van:

Just for completeless of the possibilities: you could also use AFTER INSERT triggers to store either a flag or a TableName and a PrimaryKey of the newly inserted/updated row in some helper table, which your SA application would poll regularly. However, be careful as inserts from SA would also trigger the trigger. – van