What it is and why

LowLatencyPubSub (LLPS) is a network for fast message delivery from writers to listeners over a distributed infrastructure, instead of direct client-to-client links.

Message model:

• Messages up to ~64 KB (like datagrams, not a byte stream).
• No persistence: fire-and-forget; at most a short local buffer.
• No strict ordering: messages may arrive in any order.
• Logical channels: you publish and subscribe by tenant and channel name.

Benefits:

• Speed — Delivery uses an overlay network that picks fast paths. Sometimes the direct path is slower or blocked (e.g. by firewalls or DPI); the overlay can be faster.
• Simplicity — You connect once (WebSocket or gRPC), use a token, and publish or subscribe by channel. No need to manage many peer connections.
• Fanout — One writer can reach many listeners without opening a connection to each one. The service handles distribution.

Where to go next

• Delivery semantics — what to expect (ordering, duplicates, drops, propagation delay)
• Security & Auth — tokens, permissions, best practices
• Operations — observability, limits, scaling