Syself Autopilot: GPU Kubernetes for LanguageTool Compliance
Introduction
LanguageTooler GmbH is a SaaS company based in Germany. It maintains the open-source project LanguageTool, an AI-powered writing assistant that checks grammar, spelling, style, and punctuation in multiple languages. LanguageTool improves millions of texts every day.
The Challenge
As LanguageTool grew, so did their infrastructure. They used multiple cloud providers, but most of their infrastructure was self-managed Hetzner bare metal servers.
They ran Kubernetes clusters with MicroK8s on some of these servers and used Ansible to manage workloads on the servers outside the clusters. This setup didn't scale well and wasn't reliable enough. Their 250 Hetzner bare metal servers with GPUs took a lot of management work, which was difficult to handle in-house.
Running a self-managed system also meant they were responsible for all maintenance, bug fixing, and firefighting. That took a lot of attention away from their core business, because the team spent its time maintaining infrastructure when it wanted to build the product.
Millions of people use LanguageTool every day, so the system has to be powerful, reliable, and scalable. They looked for a solution that met those demands on their cloud infrastructure and reduced the maintenance work they had to do in-house.
The Solution
They compared offerings on several public clouds and chose Syself Autopilot. The main reason was Syself's software-based approach, which automates all relevant tasks of maintaining Kubernetes clusters.
Syself could also connect to LanguageTool's existing Hetzner infrastructure. They kept Hetzner's affordable servers and got a Kubernetes solution in the same league as the main players in the market.
Syself integrated all GPU servers tightly into the lifecycle of the cluster. Because operations are automated and optimized, LanguageTool spends little time maintaining their infrastructure.
When an issue comes up, our experts are available to help. That takes pressure off their internal DevOps team, which can spend more time on its many application-related tasks.
LanguageTool has very strict compliance requirements, so automating Kubernetes compliance was essential. They chose a dedicated management platform that runs on their own infrastructure. They don't share it with anyone, and they fully control who can access it. Syself ships regular updates to keep the platform up to date.
Results
A highly automated, reliable platform now runs LanguageTool's Kubernetes clusters. The clusters include 250 Hetzner bare metal servers with GPUs and receive millions of requests a day, with peaks of around 20,000 requests per second. That shows how Syself's enterprise Kubernetes solutions scale to the most demanding workloads.
They have significantly reduced the work of maintaining their infrastructure and can now concentrate fully on their core business.
