Glossary
Concurrency (concurrent users)
Concurrency, in the context of real-time platforms, is the number of users or connections active simultaneously, and it is the primary driver of infrastructure sizing and cost.
General definition
Concurrency describes how many clients are connected and sending or receiving data at the same moment. It differs from total registered users (MAU) and even from peak sessions in a day (DAU): a platform might have one million registered users but only ten thousand online simultaneously during its busiest hour.
- Peak concurrent users (PCU): the highest simultaneous connection count ever observed
- Average concurrent users (ACU): the mean across a time window, used for capacity planning
- Connection-level concurrency: open WebSocket or XMPP connections held by the server
- Request-level concurrency: simultaneous in-flight API requests at any instant
Concurrency directly determines server memory, connection-pool sizing, and the throughput budget. Real-time protocols like WebSocket and XMPP maintain persistent connections per user, which makes concurrency management more complex than stateless HTTP APIs.
In the Ethora ecosystem
On a cloud-hosted Ethora plan, concurrency limits are baked into each pricing tier. The provider’s shared infrastructure handles connection scaling transparently, and the SLA guarantees performance up to the contracted concurrent-user ceiling. Customers check the pricing page to match their expected peak to the right tier.
With a dedicated or self-hosted chat server, you set your own concurrency target and provision accordingly. Ethora’s ejabberd-based backend supports clustering to distribute connections across nodes, and the team can help you run load tests to validate your configuration at peak concurrency before launch. In regulated environments this also lets you cap and audit concurrent access at the infrastructure level, an advantage for compliance.
