LinkedIn
The enterprise architect's guide to the service mesh
Download ebook
Forsta provides a global market research and experience management platform used by organizations across many industries to collect, analyze, visualize, and act on customer and employee data. In addition, Forsta's engineering organization is structured around individual product teams, supported by a central platform engineering function (the team interviewed for this case study). Rather than owning a single product, platform engineering supports infrastructure across the business, with development distributed across international teams and regions.
The platform engineering team under Forsta first adopted Linkerd. Forsta was acquired by Press Ganey, a trusted analytics platform in the US healthcare industry, and then underwent an acquisition a second time when Qualtrics, a leader in experience management technology, acquired Press Ganey in 2026. The platform engineering team still operates under the Forsta’s technology stack.
Forsta’s platform runs 40+ Kubernetes clusters on Azure with 2000+ pods on each of them. Some of these clusters are for R&D but most of them carry production workloads across ~400 services in total. The platform engineering team owns the Linkerd service mesh and supports the other product teams. The platform manages both Linux and Windows workloads in Kubernetes and currently has Linkerd covering their Linux workloads. Linkerd runs alongside a familiar cloud native stack that includes Argo CD, cert-manager, Grafana, and Envoy Gateway.
Early adoption of open source Linkerd
Before adopting a service mesh Forsta’s internal service-to-service communication ran over unencrypted HTTP, and load balancing relied entirely on default Kubernetes behavior. As the platform scaled, the team identified three core requirements that a service mesh needed to solve:
- Mutual TLS to encrypt service-to-service traffic
- True load balancing beyond what Kubernetes provides out of the box
- Routing rules to manage traffic as the service architecture grew in complexity
The team evaluated the Istio and Linkerd, the two most popular projects in the service mesh landscape and selected Linkerd because of its focus on operational simplicity. The team originally started out with the open source Linkerd and were able to deploy it in just a few days. Linkerd went into the platform templates and patterns that platform engineering maintains, so application teams inherited the mesh rather than adopting it. Developers did not install it, configure it, or reason about it. A team could just ship a service using the standard template and it arrived meshed.
“We auto-deployed it without too much effort and still use it without our developers needing to think too much about it”
- Jan Lønsetteig, Distinguished Software Engineer
Transition to Buoyant Enterprise for Linkerd (BEL)
Forsta ran open source Linkerd in production for multiple years and successfully managed the traffic across the ~40 clusters with minimal operational overhead. However, the open source project relied on the community forum for resolving issues which provided no guaranteed response times and had also shifted focus to edge releases. This became an increasingly untenable position for infrastructure that several hundred internal applications now depended on. Therefore, Forsta decided to upgrade to the Buoyant Enterprise for Linkerd (BEL).
By upgrading to BEL, they now had access to enterprise-grade, supported releases and a known release cadence which made planning upgrades much more predictable. Additionally, they now had a direct support line to the engineers that created Linkerd and a designated customer success manager for strategic planning. Most importantly, the move from open source Linkerd to BEL did not require them to change anything within their technology stack.
Today, BEL is a critical component of its internal developer platform and something their developers never have to think about. When the platform team walks a new parent company or an auditor through its internal developer platform, Linkerd is part of what it points at as evidence that the stack is modern and built for cloud. Asked whether BEL changed how the team operates day to day, whether it meant more confidence in deployments, the answer was that it changed none of those things, because there was nothing to change.
"It is totally transparent to the rest of the organization and very painless. It just works."
Hans Olav Damskog, Vice President - Technology
Over the years, it has also helped accelerate compliance for important certifications like SOC2, IRAP etc. Since Linkerd manages mTLS centrally via the platform, every new service automatically benefits from encryption out of the gate. From a compliance standpoint, this is a much simpler proposition than collecting individual attestations from 400+ services.
BEL operating cost: 0.1% of a headcount
From implementation to ongoing maintenance, BEL has remained a highly reliable solution with incredibly low operating costs. BEL has allowed Forsta to continue operating with a very lean platform team and today, attributes barely 0.1% of their operational headcount towards managing the BEL deployment.
Network security and traffic resiliency were the problems that first brought the Forsta platform team to adopt a Linkerd open source years ago. What has kept them with BEL in their stack since then is the low total cost of ownership (TCO) and zero-config experience for application developers.
For a lean platform team, this has become a compounding benefit. A core platform component has absorbed years of organization change without demanding attention or headcount. Asked what they would tell another platform team evaluating service meshes:
“It is a key infrastructure that we never have to be concerned about and has worked well over many years. That is a good sign.”
- Jan Lønsetteig, Distinguished Software Engineer
Interested in what Buoyant Enterprise for Linkerd can do for your team? Contact us to learn more.