2 Attribution of the quote
source | link

Running your queries in parallel from your application, and then combining the output will never be as fast as running a tuned query from the database.

Do the work where it is most appropriateDo the work where it is most appropriate

Database engines are designed and optimized to work with data sets. Certainly some database engines are more efficient at this than others, but this is their primary task. Just by using a foreign key reference, your database can increase the performance connecting two tables (JOINing them) by multiple orders of magnitude.

Even when you can show me something that executes faster on the client than the database, you will still have to factor in the time it takes to move the data down the network pipe. It is naive to move 100K records to the client just to filter it down to the 1 or 2 that you need.

Sending large sets of data across the network will also negatively affect performance of other applications that use the network.

Running your queries in parallel from your application, and then combining the output will never be as fast as running a tuned query from the database.

Do the work where it is most appropriate

Database engines are designed and optimized to work with data sets. Certainly some database engines are more efficient at this than others, but this is their primary task. Just by using a foreign key reference, your database can increase the performance connecting two tables (JOINing them) by multiple orders of magnitude.

Even when you can show me something that executes faster on the client than the database, you will still have to factor in the time it takes to move the data down the network pipe. It is naive to move 100K records to the client just to filter it down to the 1 or 2 that you need.

Sending large sets of data across the network will also negatively affect performance of other applications that use the network.

Running your queries in parallel from your application, and then combining the output will never be as fast as running a tuned query from the database.

Do the work where it is most appropriate

Database engines are designed and optimized to work with data sets. Certainly some database engines are more efficient at this than others, but this is their primary task. Just by using a foreign key reference, your database can increase the performance connecting two tables (JOINing them) by multiple orders of magnitude.

Even when you can show me something that executes faster on the client than the database, you will still have to factor in the time it takes to move the data down the network pipe. It is naive to move 100K records to the client just to filter it down to the 1 or 2 that you need.

Sending large sets of data across the network will also negatively affect performance of other applications that use the network.

1
source | link

Running your queries in parallel from your application, and then combining the output will never be as fast as running a tuned query from the database.

Do the work where it is most appropriate

Database engines are designed and optimized to work with data sets. Certainly some database engines are more efficient at this than others, but this is their primary task. Just by using a foreign key reference, your database can increase the performance connecting two tables (JOINing them) by multiple orders of magnitude.

Even when you can show me something that executes faster on the client than the database, you will still have to factor in the time it takes to move the data down the network pipe. It is naive to move 100K records to the client just to filter it down to the 1 or 2 that you need.

Sending large sets of data across the network will also negatively affect performance of other applications that use the network.