Question

I have written a program that use one single connection with prepared statements. So far this works but I’m moving into using pooled connections.

I’m making the assumption that prepared statements are used together with connection pooling but I can’t get me head around how it is used in practice.

I assume that rather than having the connection object stored I create a new one for every query I run(which will be pooled).

Before I used something like this:

//Single connection at startup
var conn = NpgsqlCommand(...);
...
//Prepare all commands once at start-up
var cmd = new NpgsqlCommand("INSERT...", conn);
cmd.Parameters.Add("id", NpgsqlDbType.Integer);
cmd.Prepare();
...
//Using the command multiple times later on
cmd.Parameters["id"].Value = 4312;
cmd.ExecuteNonQuery();

Now using pooled connections, how do I prepare these commands?
I guess I would need one prepared command for every connection, but I would also need some way to get the prepared command given a connection which could be anyone from the pool.

I assume the pooling also can create new connections, then I would need to prepare a new set of commands or do I create a new NpgsqlCommand, prepare it and execute it, for every call?

Some background:
My current setup is a test with a single user where only a few calls, all of them in sequence. Later on this will be a multi-user/multiple-thread service with parallel execution.

I can see how I could create my own connection pool and write my own code to prepare commands for new threads but I have an assumption that I won’t have to write this, but I might be wrong.

I’m using Npgsql(PostgreSQL) but if possible a general solution is appreciated.

Answer 1

According to this question the answers are that you can’t use the connection pool.