What is the benefits of using connect_lazy over connect in PgPoolOptions

[thangchung]

I have searched around in this repo and the internet to see what is the difference between connect_lazy and connect functions when using sqlx. I found the docs at https://docs.rs/sqlx/0.4.1/sqlx/pool/struct.PoolOptions.html. Is it quite clear to me.

connect: Creates a new pool from this configuration and immediately establishes one connection.
connect_lazy: Creates a new pool from this configuration and will establish a connection as the pool starts to be used.

But I just don't convince myself that what is the benefit of using connect_lazy over connect functions? Because when used with connect_lazy, we might sacrifice the async/await model, and it might not good for the intensive I/O application type. Is it right?

When uses with connect_lazy, the code looks like

PgPoolOptions::new()
        .max_connections(DB_POOL_MAX_CONNECTIONS)
        .connect_timeout(std::time::Duration::from_secs(10))
        .connect_lazy(&db_uri)

But when used with connect, it has async by the end like

PgPoolOptions::new()
        .max_connections(DB_POOL_MAX_CONNECTIONS)
        .connect_timeout(std::time::Duration::from_secs(10))
        .connect(&db_uri)
        .await

Could anyone explain a bit about this for us? Thank you very much.