播放框架+ JDBC +期货

Question

Assuming I obtain a JDBC connection through injection, like so:

class SqlQuery  @Inject()(db: Database) extends Controller  { /* .... */ }

And that the pool of connections is large enough, for example 100. Is it possible to create a Future to avoid blocking when running the SQL statement (similar to Slick futures)? Or the fact that the number of connections in the pool is large means that the SQL statement will not block?

Answer 1

Using futures is not synonymous with non-blocking. Futures allow you to execute code on another thread, or some type of executor, in general. However, the code you execute can still block.

JDBC is a blocking API. This means that when you execute a query through JDBC, the calling thread is blocked while it waits for a response from the database. Another term for this would be synchronous . A non-blocking or asynchronous API would accept a response asynchronously, freeing the calling thread from actively waiting for it. Reactive slick uses it's own driver to accept responses from a database in an asynchronous manner, which means the calling thread can be freed as soon as the query is dispatched to the database.

The difference between the two is this:

Imagine your application has a database connection pool of size 100, and a fixed thread pool of size 10. Then, let's say you wrap all of your JDBC calls in futures. Let's also say that your SqlQuery controller has a method that makes several JDBC calls at the same time. All of these queries will be run in parallel, until the thread pool is exhausted, which means you would only be able to run 10 queries at the same time at any given moment. While the calling thread would not be blocked by the JDBC calls, the threads executing them would. With enough queries running in parallel, the thread pool would become exhausted and it would no longer matter how many connections were in the pool. You could deal with this by making your thread pool larger, or using a fork join pool that expands as needed, but this could incur performance costs due to the creation of new threads and context switching. After all, your CPU is limited.

Using an asynchronous database driver like reactive slick would not block your limited pool of threads, and you would be able to run as many queries concurrently as you had connections in the pool (100 in this example). Saving threads from being blocked means saving CPU time that would otherwise be spent just waiting for responses, which means you can use it to continue to handle other requests, etc.