Realm connection utilization in Atlas is generally low, but Realm requests can sometimes cause connection spikes.

For example, after a Realm webhook is executed, Realm will open one or more connections to your cluster and ask the driver to keep its own pool of connections open. This means when the webhook hits N number of servers, you can expect >= N connections open to your Atlas cluster. N is subject to change and is related to the number of Realm servers we have in different regions and whether your Realm app is locally or globally deployed.

If you have a high number of concurrent requests, the number of connections increase in Atlas and will be reused whenever possible. New connections are opened and pooled indefinitely by the driver, and the driver will choose when to close the extra connections.

Before moving a Realm app to production, we recommend testing your configuration in a staging Atlas environment by simulating your workload and webhook requests to ensure that your current Atlas instance size is sufficient to handle the Realm workload.

Did this answer your question?