From a garage in 1999 to the base layer of global communications, Asterisk is the hardest piece of open-source real-time communication to route around. It never built a pretty interface — and almost every open PBX, call centre, IVR and voice platform still ends up back on it.
If you have ever used a free office phone system, dialled through a company's voice menu, or waited in a support queue, there is a good chance Asterisk was running underneath. It is a software PBX that runs on ordinary servers and implements voice, video, messaging, conferencing and fax entirely in software, with no dedicated switch hardware required.
Asterisk describes itself as an open-source communications framework rather than a plain telephone switch, and the distinction matters. A switch only connects and disconnects calls. Asterisk breaks every call into events that a program can control — who is calling, which queue they landed in, how long it rang, who picked up, whether the call is recorded, whether it goes to voicemail — all of it can be driven from code.
Under the hood there are four layers: channel drivers that handle signalling (chan_pjsip is the workhorse today), codec and bridging modules that handle media, the dialplan that carries business logic, and the interfaces that expose all of it to the outside world (AMI, ARI, AGI). Stack those four together and you get a communications middleware you can reshape at will.
Asterisk's real value is not that it can make calls. It is that a call becomes something you can write as code.
In 1999 Mark Spencer wrote the first version of Asterisk because he needed a phone system for his own company and found commercial PBX quotes absurd. He then founded Digium and commercialised the software — maintaining the community edition while selling hardware cards and support. That open-core plus commercial-services structure became a template for countless infrastructure projects since.
In 2018 Digium was acquired by the Canadian communications vendor Sangoma, and maintenance moved with it. Sangoma is still the primary enterprise maintainer, while contributors come from carriers, integrators and independent developers around the world. The project has kept a steady rhythm: one major release a year, even-numbered releases becoming long-term support, odd-numbered releases serving as standard releases.
The policy is fairly clean: a new major version lands around October each year. Even numbers enter the LTS track with roughly four years of full maintenance followed by a year of security fixes only. Odd numbers are standard releases with a support window of about thirteen months, used mainly to trial new features.
For production systems that rhythm means an upgrade decision is roughly a one-year project: an expiring LTS is not a 'deal with it later' item, but a full engineering cycle of recompiling, regression testing, validating call paths and confirming third-party module compatibility.
There is no shortage of more modern options: dedicated media servers for WebRTC, turnkey SaaS for cloud contact centres. Asterisk has held its ground for three reasons.
First, malleability. From the dialplan down to the ARI interface, nearly every part can be swapped out. Carriers needing unusual billing, enterprises wiring telephony into internal ticketing, hardware vendors building custom endpoints — all of them eventually need that level of programmability.
Second, protocol coverage. SIP, PJSIP, IAX2, WebRTC and legacy TDM interfaces coexist, so it can face new endpoints and old equipment at the same time. Most enterprises are half new and half old, and that mixed estate is exactly what pure-cloud offerings handle worst.
Third, the cost structure. The software is free, the hardware is commodity servers, and cost lands on integration and operations. At high call volumes, with a technical team in place, the maths usually beats per-seat licensing by a wide margin.
Small and mid-sized businesses use graphical distributions such as FreePBX to stand up an office phone system quickly. Call centres and outsourced support teams build their whole operation on queues, recording, reporting and CRM integration. Carriers and internet telephony providers use Asterisk for access and application layers. Hardware and solution vendors embed it in intercoms, door entry, dispatch and public-address products.
Those four groups want very different things: businesses want stability and easy maintenance, call centres want concurrency and reporting, carriers want protocol interoperability and billing, vendors want something they can trim down and embed. Asterisk serves all of them because of a thin core with thick extensions.
If this is your first contact, get one call working before anything else: install an LTS release, configure two SIP extensions, call between two softphones, add a dialplan rule to transfer, then build a small IVR. That loop teaches you the channel, context, priority, application model better than any manual.
Production adds three concerns. Media paths come first — NAT and firewalls are where newcomers lose the most time. Then capacity planning: transcoding is CPU-hungry, so avoid it where you can. Finally observability: get logs, CDRs and AMI events out of the box before you need them.
As for whether to run LTS, the answer is almost always yes — unless you are prototyping, there is little reason to accept a shorter support window for the sake of new features.