What is SIP trunking?
SIP trunking is how many business phone systems connect to the outside world today. A “trunk” is a connection that carries many calls at once between a phone system and a carrier. In the past, trunks were physical lines or digital circuits. A SIP trunk does the same job over an IP data connection, using the Session Initiation Protocol (SIP) to set up and manage each call.
Calls travel as Voice over IP (VoIP) between the organisation and its carrier. The carrier then connects them to the public switched telephone network (PSTN) and mobile networks, so customers can ring in from any phone and agents can ring out to any number.
How SIP trunks work
A SIP trunk has a few key elements:
- Channels. Each channel carries one simultaneous call. A trunk might have 10, 100 or thousands of channels, depending on need.
- Numbers. Phone numbers, including DDI numbers and freephone numbers, are pointed at the trunk so calls to them arrive at the right system.
- The phone system. An on-premise PBX, a cloud phone system, or a contact centre platform receives the calls and routes them.
- Security and redundancy. Encryption, firewalls and failover paths protect the service and keep it running if one link fails.
With fully cloud-based CCaaS and UCaaS platforms, the provider often supplies the telephony connection as part of the service, so the customer never sees a trunk at all.
Why SIP trunking matters
Compared with traditional lines, SIP trunks are usually cheaper, quicker to scale and more flexible. Channels can be added without installing new lines, numbers can be moved between sites, and calls can be redirected in software if a site goes offline. That flexibility supports business continuity and makes it practical to have agents working from many locations.
Capacity: the hidden limit in a call spike
For contact centres, the most important SIP trunking question is capacity. Every caller uses a channel from the moment the call connects, including while they wait in queue or listen to an IVR message. In a call avalanche, the number of simultaneous callers can rise far beyond the number of agents. If the trunk runs out of channels, new callers hear an engaged tone or the call fails before it reaches a queue. They don’t show up in queue statistics, so the problem can be invisible in reports.
Good practice includes:
- Sizing for peaks, not averages, including queued and IVR callers.
- Agreeing burst capacity with your carrier or platform provider for emergencies.
- Monitoring channel usage and alerting when it nears the limit.
- Building in failover to a second path or provider.
- Answering more callers instantly with self-service or an AI voice assistant, so calls resolve and free their channel sooner.
Things to check with a provider
Ask how many concurrent calls are included, how quickly capacity can be raised, what happens when the limit is hit, how calls fail over during an outage, and where call data and recordings are hosted.