Nexcess.com Servers.com LiquidWeb.com
Back

Customer Story

Sipfront keeps critical phone systems working, with test capacity custom-built in hours

Sipfront provides automated quality assurance for telecom systems: carrier networks, contact centers, and emergency phone lines that have to work every time. Testing these systems means generating enormous volumes of call traffic on demand, a workload most hosting providers would misread as an attack, and one few can even support at all. Sipfront needed a partner that could custom-build server capacity for exactly that job and deliver it in hours.

SaaS 5 min read
Sipfront

At a glance

Andreas Granig has stood in a data center at 2am, testing a software upgrade case by case, two phones in hand, blowing into one to check that the other side hears it, racing a 6am deadline before real callers flood the system. He knows the risks of what can happen when exhausted engineers forgo this full QA testing effort. If the updated system belongs to emergency services, for example, the cost of even a single missed test can be measured in lives.

Andreas Granig, CEO and Co-Founder of Sipfront

"The age of manual testing has to be over."

Andreas Granig

CEO and Co-Founder of Sipfront

Andreas Granig, spent years manually testing software upgrades to telecommunication systems. He was often running QA throughout the night to ensure that the new software would work as required, since emergency services companies were dependent upon the quality and uptime of their telecom network. He was convinced that there had to be a better way to test new software.

That conviction became Sipfront, the company he founded in 2020 after 25 years spent building the plumbing of modern telephony, including open source software that now runs inside VoIP systems worldwide, including at AWS Connect and at Google. "If you've made a phone call today, chances are it already ran over software we developed."

Today Sipfront tests it all automatically: routine functionality, crushing concurrent call loads, end-to-end call quality, and, since 2026, voice AI agents now answering many of those calls. Their tests help businesses address scenarios they fear most, sudden spikes in demand for service. When a surprise storm threatens to leave rail passengers stranded and cause thousands of anxious callers to hit a voice bot at once, Sipfront proves the system will hold before the storm arrives. When a large German telco moved its contact center onto new core systems, Sipfront and its integration partners verified that 5,000 contact center agents could go online at 7am the morning after the migration.

The challenge

Sipfront started on AWS deliberately, to avoid managing servers while the product found its footing, knowing the economics of simulating tens of thousands of calls on a hyperscaler solution would prove costly as the company grew. Once bills exceeded pain tolerances, they sought out a more cost-effective approach. Their first bare-metal provider offered no capacity in the US or Southeast Asia, where customers needed test traffic to originate, and the vendor’s automated security systems kept mistaking Sipfront's legitimate test traffic for an attack. Nobody there took the time to learn the difference, and other hosts were worse. Granig said, "One provider in the US just off-boarded us due to our load. When we opened support tickets, they were like, we can't do anything about that."

Sipfront needed a partner that would build for its workload instead of flagging it as an attack, put capacity wherever its customers needed it next, and answer a support call when it mattered.

The solution

Servers.com by Nexcess custom-builds Sipfront's servers for the workload's exact demands and delivers them at project speed. "We get custom-built servers in a matter of hours mostly, and sometimes days depending on the availability in the region, but they are very much tailored to our workload," Granig says.

The flexibility of this bespoke arrangement unlocks a key benefit for Sipfront. Its control plane and orchestration stay on AWS, running exactly as before, while the heavy test traffic runs on Servers.com bare metal. The team kept the workflow it already had and moved only the workload that was breaking the budget.

"Servers.com understands what we do. They understand our traffic patterns, and they can accommodate them," Granig says. The relationship is a partnership, with dedicated support channels and regular check-ins designed to anticipate needs rather than respond to them. As Sipfront's testing extended into voice AI, the same partnership supplied GPU servers for the specialized quality-analysis models it runs in-house.

Interested in the engineering story behind this partnership, including the traffic pattern most providers mistake for an attack? Read the technical companion.

To learn how Servers.com by Nexcess builds bare metal cloud for workloads like Sipfront's, contact the team.

Thanks to Andreas Granig for sharing the Sipfront x servers.com story.

Find out which part of your platform belongs on bare metal

In a workload placement review, a solutions architect maps your current setup, shows which workloads to move first and which to leave, and models what the same workload costs over the next 12 to 36 months at your current growth.