Thanks to Slava for arranging this meeting. Vance Shipley is the founder and CEO of Sigscale. Sigscale delivers signaling and distributed fault tolerant systems. They have developed their protocol stacks (SIGTRAN, TCAP, CAP, MAP, NGAP, RADIUS, EAP) in Erlang, allowing them to build microservices with very low footprint and massive scalability.

The High School Drop, paying $10k for an Erlang License

Beginning in the beginning, Vance dropped out of high school and started working in the cable industry at the age of 16, which was adding telephony to their offer. He compared the telephony service to a finite state machine.

And so began Vance’s a 25 year journey in Erlang and the Open Telephony Platform. He began with Motivity Telecom, a software developer serving the communications service provider (CSP) market.

He developed a product called the Flexible Signalling Gateway (FSG) which provided an IWF (Interworking Function) within and between SS7 ISUP variants and ISDN PRI. Examples of it’s application include; interworking ANSI and ITU ISUP, aggregating PRI into ISUP and mapping and filtering individual ISUP parameters.

He needed concurrency, and hence discovered Ericsson had a technology called Erlang and the Open Telephony platform. He bought one license for $10k, and it came with a stack of hardback books. 6 months later they released Erlang under an open source license. Vance decided to be happy about shift to open source as he’d never need to spend the $10k again.

Interworking Functions

His first customer was a carrier that needed to generate lots of traffic, to meet regulatory requirements on generating and much traffic as they terminated. This story is a classic Erlang one. Vance ran his app, everything appeared to be working, traffic was generated, but the error logs were vast. The app was crashing on call teardown. But the Erlang supervisor managed that, so the app ran, the customer was happy, and Vance discovered his lifetime programming language, Erlang.

Slava asked Vance about Sigscale and the services provide to CSPs. The body of work Vance had built included the telecom signalling stack, and running in Erlang enables signalling at scale. His rationale for using open source was purely economics. Avoid raising venture cash, publish on github and people will discover there’s an open source version of what they are paying cash for. It worked!

Power of Open Source, never making a cold call

He’s never made a cold call, people call Vance. He open sourced the full SS7 stack, and the acronym soup on top, SIGTRAN, TCAP, CAP, MAP, NGAP, RADIUS, EAP, and the ASN compiler. ASN.1 is used within SS7 protocols, specifically within the Transaction Capabilities Application Part (TCAP), to define the format of messages exchanged between network elements for advanced services.

Vance paid tribute to the ETSI and 3GPP standards written for SS7/TCAP that made his life much easier. I asked if he was part of the standards process, as standards like IMS and RCS are far easier to implement being on the inside.

Vance made clear, their clarity was important, they were not describing defacto interfaces. Though in my experience, knowledge on how Nokia, Ericsson, Lucent implemented SS7/TCAP was important, which led to Vance creating interworking functions at Motivity, particularly for international signalling. Or enabling ANSI switches to work directly with ETSI switches to avoid to avoid the cost of international gateways.

Extending Erlang with Erlang-RED and Elixir

I then asked about interesting Erlang projects, Vance mentioned Erlang Red, that we are covering next week with Gerrit Riessen, Experimentalist FlowHub.

Erlang runs in a virtual machine, the focus has been extending Erlang to other languages like Elixir and Lisp. Erlang and Elixir are two programming languages that run on the same virtual machine, the (BEAM), the virtual machine at the core of the Erlang Open Telecom Platform (OTP), providing a foundation for highly concurrent, distributed, and fault-tolerant systems.

Slava highlighted that Elixir work was created for web applications / websites. The Financial Times and Discord run on Elixir. Newcomers to BEAM, tend to focus on Elixir rather than Erlang, because of their interest in the web. I find myself explaining that fact to many in the telecom industry. The web won, developers focus on the web to get jobs. However, building with Erlang gives the best native performance.

Jonathan Eisenzopf asked about AI use cases Erlang users in telecom are evaluating. And Vance answer was, ‘not a lot’. His concern was on the efficiency of AI and the resources used, Erlang is efficient in resource usage.

Slava sees Erlang providing the reliable connectivity to AI applications. I raised the issue with spam, scams, and robocalling, and whether AI has a role there. But it’s clear this is not Erlang’s wheel house, a concurrency oriented functional programming language. Vance reviews the resource required for OCS (online charging system), its uses GCP (Google Cloud Platform) smallest virtual machine, a few dollars per month for redundancy.

Slava’s Final Questions

We wrapped up on Slava’s final questions. What open source projects are missing? Vance mentioned 5GSA, but highlight the open source solutions would not be used by a ‘serious provider,’ Ericsson / Nokia. They tend to be used in the labs or smaller applications. But history shows those prototypes evolve into production systems. And Erlang can re-invent to 5GSA core cost.

Slava’s last question was to name the carriers using Erlang. Vance started with WhatsApp is built on Erlang. Web-centric companies have a long history on naming open source projects in their infrastructure.

Ericsson’s ATM switch core was Erlang. BT and O2 have deployed Erlang, O2 has a respected Erlang dev team. Vodafone NZ SMS infrastructure was based on Erlang. The cultural issue is it’s frowned upon to name open source infrastructure. On the web its proudly stated as a mark to be proud of. This is a hot topic for me and TADSummit. Open source is important to the telecom industry and its future, we need to change the culture on talking about open source.