How the network works

1BSV is a layered multicast network: a transaction is sent once and the fabric carries it to everyone who subscribed — numbered, repairable, and built to grow without a ceiling.

One transaction, end to end

A sender hands its transaction to the nearest entry point. From there the network does the work — once.

Pipeline: sender to ingress to multicast fabric to listeners and cache to consumers
sender → ingress → multicast fabric → listeners + cache → consumers
Ingress

Stamp, sort, send once

Accepts a transaction, numbers it, sorts it into a delivery lane by a fixed rule, and broadcasts it a single time.

Fabric

The shared lanes

The multicast fabric carries each lane to everyone subscribed to it — no per-receiver copies.

Listeners, cache & consumers

Receive, repair, deliver

Listeners tune in and watch the numbers; a cache resends anything missing. Miners and apps get exactly what they subscribed to — numbered and complete, with nothing silently lost.

What the network does

Four properties carry the whole design.

01

Ingress points scaled widely

Entry points sit close to senders and grow in number as the network grows. A sender always hands off to the nearest door, and adding more doors adds capacity — there is no single front gate to congest.

02

Self-repairing delivery

Every stream is numbered, so a missing message is detected the instant it's skipped — never guessed. The receiver asks a nearby cache to resend it; if that cache doesn't have it, the request moves on immediately. Loss can't hide, and it can't persist.

A numbered stream with a gap at position three; a request is sent and the missing message is resent, producing a complete in-order stream
a gap is detected and the missing frame is recovered — nothing stays lost
03

Sharded and filtered delivery at the edge

Traffic is split across independent lanes by a fixed, verifiable rule — the same transaction always takes the same lane, and anyone can recompute it. Receivers tune in to only the lanes they care about, so bandwidth tracks interest, not the size of the network.

A transaction is mapped by a fixed rule onto one of several delivery lanes
receivers narrow further at the edge · subscribe to subtrees to take only the slice you need
04

Highly available delivery

Receivers connect over the path that fits them — redundant tunnels or dedicated cross-connects — with automatic failover. If a path drops, delivery continues on another without losing the stream.

Built to scale horizontally

Multicast is the one medium that gets cheaper per subscriber as it grows: publish once, reach everyone tuned in.

Growth is simple — add network, add compute. There's no central bottleneck to widen and nothing to re-architect; capacity spreads across independent lanes and ingress points. The path runs toward 1B+ transactions per second.

The protocol is open

Every mechanism here is published as an open standard anyone can implement and verify.