Why Asynchronous Database Connectivity Matters: Rust and SQLite ORM

When it comes to database connectivity, many developers assume that using an asynchronous library will automatically lead to better performance. However, this is not always the case, especially when dealing with large-scale deployments.

In a web backend, there are only a finite number of hardware threads available to process requests. If these threads are blocked waiting for database responses, it can lead to wasted time and resources. This is particularly problematic if the database server is located on a separate physical computer, as is common in larger scale deployments.

To illustrate this point, let's take a look at some benchmarks comparing Actix with Diesel against Actix using async DB connectivity via Tokio Postgres. The results show that Actix implementations using async DB connectivity significantly outperform those using Diesel, demonstrating the importance of asynchronous database connectivity.

In fact, any blocking library can prevent this from happening, as it will waste time waiting for the database server to respond. This wasted time could be spent processing other requests, which is particularly important in a web backend where capacity is often reached quickly.

Of course, there are valid use cases for async-io and even async database io, but my point is that most people likely don't need it. In many cases, other factors will dominate the performance of an application, making optimizing only the IO part unnecessary.

Some developers claim that asynchronous connectivity is like 'magic fairy dust' that makes their application faster, which is simply not true. In reality, non-async implementations can perform just as well as async variants in many cases. For example, Postgres does fine in benchmarks even though it uses an internal async connection.

In conclusion, when it comes to database connectivity, asynchronous connectivity is important for large-scale deployments where capacity is a concern. However, in many cases, other factors will dominate performance, making optimizing only the IO part unnecessary. By understanding these limitations and choosing the right approach for your specific use case, you can build more efficient and scalable applications.

Rust sqlite orm