Skip to content
Phone Guides

Asterisk routing across multiple sites

Design a multi-site system around explicit destinations and permissions. A connection between PBXs makes another route possible; it does not establish that every extension at either site should inherit the other site's outbound calling rights.

Source-reviewed, 19 October 2026. Prepared from the cited documentation. No multi-site topology has been independently tested by Phone Guides. Scope: Asterisk 22/PJSIP design; carrier, endpoint and network arrangements require their own evidence.

Choose where each responsibility lives

Record which system owns phone registrations, internal numbering, incoming numbers, outbound carrier access, voicemail and management. Distinguish a central PBX serving remote phones from multiple PBXs connected by trunks. Those arrangements have different dependencies when a connection fails.

Use the official network deployment guidance alongside your network design. The PJSIP reference defines endpoint options; it does not supply a complete inter-site trust policy.

Make the numbering plan unambiguous

Item Decision to record
Internal numbers Unique numbers or an explicit site prefix
Incoming numbers Which PBX receives each carrier destination
Outbound numbers Which site/carrier may carry each destination class
Identity Which caller identity is retained or translated
Exceptions Emergency and restricted destination arrangements
Unreachable site Defined caller treatment and permitted alternative

Normalise numbers at a documented boundary. Keep the original received value available in restricted diagnostic evidence where appropriate. Avoid routing the same unknown number back and forth between systems until a hop limit ends the call.

For a small two-site plan, write the allowed ranges as an explicit matrix before creating patterns. This example is illustrative only: site A owns 2xxx, site B owns 3xxx, and each site has its own carrier route.

Call arrives at Destination Route owner If inter-site link is down
Site A 2xxx Local dialplan at A Local handling continues
Site A 3xxx A's restricted route to B A returns a defined unavailable outcome
Site A External number A's carrier policy Follow A's carrier failure route
Site B 3xxx Local dialplan at B Local handling continues
Site B 2xxx B's restricted route to A B returns a defined unavailable outcome
Site B External number B's carrier policy Follow B's carrier failure route

Do not give either inter-site peer a catch-all route to the other site's carrier. If a remote extension is unavailable, make the outcome local to the originating site instead of forwarding the same number back across the link.

Give inter-site calls a narrow context

Treat an inter-site trunk as a specific source with its own inbound context and routing rules. Permit only destinations intended for that connection. A trusted network location does not prove that every application behind it may place international calls through your carrier.

Document how the remote PBX is identified and authenticated. Keep endpoint identity separate from a caller-ID value that can vary per call. Use SIP trunk design for the object and carrier boundaries.

Draw the actual media paths

Identify whether media passes through a central PBX, stays within a site, or travels directly between endpoints. Include NAT, address translation, firewall policy, codecs and recording requirements. A signalling connection between PBXs does not prove that every endpoint can reach the negotiated audio address.

Refer to NAT and RTP and codecs and transcoding. Record encryption separately for each leg. Avoid describing a partly protected route as end-to-end encrypted.

Plan the disconnected state

For each failed link, decide which local calls, incoming routes and outbound services remain available. Check provider requirements before defining alternate routes. Do not promise continued emergency calling without the exact service and location arrangements being established.

On your own deployment, check calls in both directions, denied destinations, overlapping numbers, transfer, voicemail, identity and two-way audio. Repeat the relevant checks with the inter-site connection unavailable. Keep recovery behaviour separate from a claim that an active call survives a PBX failure.

Return to Asterisk documentation.