Measuring and improving how fast software runs and how much load it can take
View All Tags
Up to its minimum, the .NET thread pool creates threads as soon as they're needed. Past that, it adds a couple per second. So raising the minimum helps a burst of blocking work. It stopped a real crash on my API, and I measured what it costs.
When requests block threads, the .NET thread pool can't keep up, because it only adds a couple of threads a second. The app gets slow for no visible reason. dotnet-counters shows you what's happening, and here it is on a real API, right up until it crashed.
AddDbContextPool sounds like it should help with database connections. It doesn't: it reuses DbContext objects, and connections have their own pool in the driver. It's a one-word change, so I tried it on a real API, and measured no difference.
A concurrency limit lets a fixed number of requests run at once, and makes the rest wait or go away. It's different from a rate limit, which counts requests over time. It's also what kept my API alive when a burst would otherwise have crashed it.
Connect Timeout is how long MySql.Data waits for a connection, and it defaults to 15 seconds. That looks harmless until retries multiply it. An unreachable server held one request for 64 seconds in my test, and a smaller value brought that to 24.
Pooled connections can live for as long as the app runs, which is a problem if the server behind them changes. Connection Lifetime retires them once they're old enough, so new ones find the new server. I timed exactly when it happens.
After a burst, a connection pool holds a lot of open connections. How long do they stay? I opened 8, closed them, and counted them every minute: they go after about 3 minutes, down to Minimum Pool Size.
Maximum Pool Size defaults to 100, and MySQL's max_connections to 151. So two copies of an app can ask for more than the server allows, and more requests than the pool holds can end the whole process. Here's how I picked a number, from measurements.
Opening a MySQL connection takes many round trips: TCP, TLS, then signing in. So MySql.Data keeps connections open after you close them, and hands them out again. A few lines of C# let you watch it work, and see what it saves.
Three questions tell you if MySQL is near its connection limit: how many now, how many at the worst moment, and who has them. Each is one statement.
