Boosting Multi-Cloud Network Performance with NanoPing

Boosting Multi-Cloud Network Performance with NanoPing

At NanoPing, we’ve always believed in solving real-world network challenges with practical, deeply engineered solutions. In 2023, our CTO, Morten & Head of Project & Product, Peter, contributed to a project focused on a major hurdle in modern cloud environments: how to create high-performance, reliable multi-cloud connectivity between Kubernetes clusters.

This effort was originally done under our former company name, Steinwurf, and the product used was what we then called RAMP - today, that product has evolved into what we now offer as NanoPing.

The project was a collaboration with the Open 5G HyperCore initiative and focused on developing a plug-and-play solution to stitch together distributed cloud clusters. The requirement? Build a vendor-neutral, high-performance tunnel capable of running across private, public, or hybrid cloud environments, with minimal operational friction.

The Problem with Traditional Cloud Networking

When connecting Kubernetes clusters in different cloud environments, operators often hit a wall. NAT traversal issues, configuration complexity, and performance degradation are common -especially in use cases involving high-throughput, low-latency workloads such as edge computing, VR, or distributed databases.

The need was clear: enable deterministic, low-latency, high-throughput connections that “just work” regardless of whether clusters live in AWS, Azure, on-prem hardware, or something in between.

The NanoPing Approach

NanoPing (then known as RAMP) was used to build a transport-agnostic overlay network that dynamically selects the best available path, based on real-time network conditions. This included:

• FEC (Forward Error Correction) implemented using RLNC (Random Linear Network Coding) to mitigate packet loss without retransmissions.
• Ultra-low-latency design, minimizing jitter even across congested or unreliable links.
• The ability to run as a user-space application, with no kernel dependencies or root access required - ideal for cloud-native deployments.

One of the major wins from the project was demonstrating that high-performance networking can be delivered as a simple container, with zero-touch provisioning across clouds.

Results

The deployed system exceeded performance expectations in both bandwidth and latency metrics, even outperforming conventional SD-WAN or VPN setups in cross-cloud scenarios. It also provided robustness in the face of fluctuating link quality - a common reality when operating across multiple vendors and infrastructures.

The full technical deep-dive, including performance benchmarks and architectural details, can be found in the original write-up published by the Open 5G HyperCore initiative here.

Why It Matters

As multi-cloud architectures become more common, teams need tools that abstract away the messiness of cloud networking without sacrificing performance or control.

NanoPing is built precisely for this - whether it’s connecting critical infrastructure across providers or delivering real-time performance for edge applications, we provide zero-hassle, high-resilience network connectivity in a minimal software footprint.

We’re proud of the results, and even more excited about where the technology is heading.

If you’re facing similar challenges - or want to explore what high-performance, plug-and-play networking looks like - get in touch.