Posted on: 14th May 2026, by Magrathea
By Linus Surguy, Co-Founder, Magrathea Telecommunications…
In July 2026, Magrathea Telecommunications Limited reaches its 25th anniversary. For me personally, that milestone coincides with more than 30 years spent working in the telecommunications industry.
It feels like an appropriate moment to pause and reflect, not only on the growth of the company, but on the extraordinary technological transformation that has taken place since the early 1990s. Over that period, telecommunications has evolved from something physical and tangible into something largely defined by software, abstraction, and automation.
And yet, beneath all that change, many of the core engineering principles remain remarkably familiar.
The 1990s: when telecoms was physical
In the early 1990s, telecommunications was something you could see and touch. Networks were built from copper pairs, punch-down blocks, and racks of equipment designed to do one very specific job.
International exchanges were dense, noisy environments filled with patching frames, thick cable looms, and hardware whose presence alone conveyed its importance. Capacity was precious. Reliability was achieved through conservative design, duplicated components, and meticulous attention to detail.
Much of the intelligence lived in hardware. Software existed, but it was treated cautiously. If a system was stable, it was rarely changed without very good reason. One of my earliest projects took place in this environment, working on a British Telecom initiative in the early 1990s.
At the time, BT operated international satellite links to other countries, where bandwidth was scarce and extremely expensive, making efficient use of that capacity was a significant engineering challenge.
Although the network was entirely TDM-based, not all calls behaved in the same way. Speech could tolerate compression remarkably well. Fax and modem signals could not. Applying the wrong compression might allow a call to connect, but the information carried within it would fail.
BT wanted to understand what was actually being carried on its international voice circuits.
The project involved installing monitoring systems directly into BT’s international telephone exchanges. These systems analysed live TDM timeslots and attempted to classify them in real time as carrying speech, fax, or data.
There were no packets to inspect and no signalling metadata to rely on. Classification was performed by examining waveform characteristics such as tone patterns, cadence, energy levels, and timing behaviour. In effect, the network was being taught to listen.
The results were used to optimise compression equipment on satellite links. Speech circuits could be compressed more aggressively, while fax and data calls could be handled more conservatively, improving reliability without wasting precious capacity.
What made the experience particularly striking was the environment itself. Many of the exchange rooms still contained older analogue equipment that had not yet been fully removed. Large patching frames and heavy cabling stood alongside newer digital systems.
Even then, it was obvious how much smaller the TDM equipment footprint was compared to the analogue links it replaced. Standing in those rooms, you could quite literally see technological progress measured in reclaimed floor space.
It was an early lesson that telecommunications evolve not only through new capabilities, but through quiet, incremental efficiency.
When telecoms started to look like computing
By the latter half of the 1990s, the industry was already changing again.
Expansion cards had become available that could be installed into industrial PC chassis, allowing digital switching and processing of TDM signals without the need for fully custom hardware. Telecoms systems were beginning to resemble computing platforms.
One particularly interesting project from this period involved deploying a calling card system within the network of the incumbent operator in Pakistan.
Calling cards were a major service at the time, providing access to national and international calls in regions where personal telephone ownership was limited. The platform had to integrate directly into the national network, handling authentication, routing, and real-time call control.
This project introduced me to SS7 signalling for the first time.
Alongside the bearer paths ran a completely separate signalling network, invisible to subscribers but fundamental to call control. It was here that concepts such as call states, message sequencing, and network expectations became critical.
Although SS7 standards existed, the reality was far more nuanced. The operator in Pakistan used a specific variant of TUP, the Telephony User Part that pre-dated ISUP in many networks. Our systems had to be carefully adapted to match that implementation exactly.
Small differences mattered. Timer behaviour, optional parameters, and subtle interpretation of cause codes could determine whether calls completed cleanly or failed unexpectedly.
It was a valuable lesson in international telecommunications: standards define intent, but interoperability is earned through understanding how networks actually behave.
Building our own network
In the early 2000s, Magrathea Telecommunications Limited was launched.
For the first time, we were no longer building systems for other operators. We were building and operating a network of our own, using the experience gained from years of working inside others.
A key requirement for launching was deploying SS7 signalling on our own hardware.
Before we could connect to the UK national telephone network, our systems had to undergo detailed interconnect testing with BT at their Martlesham laboratories. These tests examined every aspect of call handling, from normal call setup to failure scenarios and abnormal releases.
Only once this testing was successfully completed were we permitted to connect live.
What was particularly striking was the scale involved. The initial capacity connected on our very first switch was already comparable with, and in some cases greater than, many earlier projects we had delivered for other operators.
There was no gradual introduction. From day one, Magrathea was operating as a national carrier with interconnects to multiple networks (most with names that have since disappeared into history).
That shift brought a fundamental change in perspective. Reliability was no longer something we designed for others. It became our responsibility, continuously and unavoidably.
Seeing the shape of what was coming
Even while establishing ourselves on traditional telephony infrastructure, it was clear that IP-based voice was approaching.
VoIP was still viewed with scepticism across much of the industry. Quality concerns remained, and carrier-grade reliability was widely questioned. But the trajectory was unmistakable.
Within a year or so of launching, Magrathea introduced the UK’s first wholesale VoIP-based service.
From the outset, it was designed differently from traditional offerings. APIs allowed customers to allocate telephone numbers programmatically and have them delivered directly to their own platforms using standard protocols such as SIP.
At a time when provisioning was still largely manual, this represented a significant shift in thinking.
Voice was no longer treated as a fixed service. It was becoming a software component that could be integrated, automated, and embedded into other systems.
When IP moved into the centre
Early VoIP implementations often treated IP as an attachment to existing switches. TDM remained the control plane, with IP acting as an interface.
As we developed new signalling software internally, we chose a different approach.
We moved the command and control of calls entirely into an IP core. Our own softswitches handled routing logic, service behaviour, and interconnect management natively within an IP environment.
TDM and IP were no longer peers. They became access methods into the same core.
This allowed SS7 and traditional trunks to interwork seamlessly with SIP-based services, while keeping the intelligence independent of bearer technology.
It also changed how we thought about resilience. Protection shifted from individual devices to distributed architecture, redundancy, and separation between signalling and media.
In practical terms, Magrathea had become one of the early operators with a true IP core.
The industry catches up
During the 2010s, the wider industry began to follow the same path.
Major networks started offering IP interconnects alongside traditional TDM. Magrathea became one of the first trial customers of BT’s IP Exchange product, working through the early stages of IP-based national interconnection.
As confidence in IP grew, long-discussed plans for PSTN switch-off moved steadily toward reality.
Once again, we were ready.
We were among the early operators to migrate to a pure IP interconnect with BT, allowing us to finally switch off the TDM infrastructure we had built, operated, and relied upon for nearly two decades.
There was something quietly significant about that moment. Those systems represented years of careful engineering, incremental expansion, and operational experience. Turning them off marked the end of an era.
Yet the network that replaced them was already vastly more capable.
Where we are today
Today, 25 years after Magrathea was founded, we operate a pure IP network carrying more traffic in a single day than our original systems likely handled in several months.
The physical footprint has shrunk dramatically. The capacity has expanded beyond anything that would have seemed realistic in the 1990s. Almost all of the network’s intelligence now exists in software.
And yet, the fundamental challenge remains unchanged.
Communications must be reliable, predictable, and trusted, regardless of how complex the underlying systems become. Whether the interface is a copper pair, a TDM circuit, a SIP trunk, or an API call, the expectation from the user is exactly the same.
As new platforms, services, and networks continue to emerge, voice keeps finding new places to integrate. Embedded into applications and workflows that bear little resemblance to traditional telephony.
We look forward to continuing that journey, using the same approach that has guided us throughout the past three decades: understanding how networks behave, respecting the lessons of earlier technologies, and evolving carefully as the industry moves forward.
Because while telecommunications will continue to reinvent itself, the principles that make it work endure.