Glossary
Throughput
Throughput is the rate at which a system successfully processes work, typically measured as messages per second, requests per second, or events per second.
General definition
Throughput describes how much useful work a system completes in a given time window. In messaging and real-time communications it is most often expressed as messages per second (MPS) or events per second (EPS). In API-heavy services the equivalent metric is requests per second (RPS). Throughput is distinct from latency: a system can have high throughput and still add noticeable delay to each individual message if the queue depth is large.
- Sustained throughput: the rate the system can maintain indefinitely without degradation
- Peak throughput: the maximum rate achievable for a short burst before errors or queuing appear
- Effective throughput: sustained throughput minus the rate of failed or dropped messages
Throughput is one of the key outputs of load testing and feeds directly into SLO definitions and SLA commitments. Bottlenecks that limit throughput include CPU saturation, database write contention, network bandwidth, and message broker concurrency limits.
In the Ethora ecosystem
On a cloud-hosted Ethora plan, throughput limits are defined by the chosen tier and published in the pricing page and SLA. The provider scales the underlying infrastructure to meet those guarantees; customers simply stay within their contracted message volume.
With a dedicated or self-hosted chat server, throughput is determined by the hardware and configuration you provision. Ethora’s platform is built on ejabberd, an XMPP server trusted for high-volume deployments, and the team can help you size nodes, tune connection pools, and run load tests to confirm your throughput targets before go-live. For regulated industries this also means you can document and audit exactly how many messages flow through your system.
