Photon Realtime 4.0: Adjusting per‑room message rate to mitigate concurrency latency

Understanding the 500 msg/s Threshold

In Photon Realtime 4.0, the 500 messages per second per-room guideline serves as a stability threshold. When the aggregate message rate of all peers in a single room exceeds this limit, the server typically begins queuing messages. This manifests as increased latency (jitter) for all participants in that room, as the server cannot process and broadcast events as quickly as they are being received.

Behavioral Impact of Exceeding Limits

The effect of exceeding this rate depends on the duration and intensity of the spike:

• Transient Spikes: Short bursts may be absorbed by server buffers, resulting in a temporary increase in round-trip time (RTT).
• Sustained Overload: Continuous traffic above the threshold leads to growing queues. This can eventually trigger packet loss or forced client disconnections to protect the health of the wider server instance.
• Concurrency Effects: When many peers send events simultaneously, the server must handle the multiplication of these events (1 sender × N recipients). High concurrency accelerates the approach to the 500 msg/s limit even if individual clients are sending infrequently.

Cloud vs. Self-Hosted Deployments

The ability to mitigate this latency differs fundamentally based on your deployment model:

Feature	Photon Cloud	Self-Hosted Server
Limit Configuration	Fixed/Managed; non-configurable.	Configurable via PhotonServer.exe.config.
Latency Mitigation	Must optimize client-side send rates.	Can increase thresholds if hardware permits.
Risk Factor	Strict enforcement/Disconnects.	CPU exhaustion/System instability.

Mitigation Steps

1. Implement Send-Rate Throttling: Use a client-side timer to ensure peers do not exceed a specific message frequency (e.g., 20Hz for movement).
2. Utilize Interest Groups: Reduce the number of recipients per message. While this lowers the network load on clients, note that the server still processes the initial incoming event.
3. (Self-Hosted Only) Adjust Config: Modify the server configuration to raise the message-limit threshold, provided you have monitored CPU overhead and confirmed available headroom.

Diagnostic Requirement: To provide a more specific recommendation, please clarify if you are seeing DisconnectCause.ServerTimeout or if the latency is observed as a gradual climb in RTT without disconnection.