Open Source Summit + Embedded Linux Conference North America 2026

Delivering predictable, high‑quality network services in dense, multi‑access edge environments remains a central challenge for operators pursuing a Network‑as‑a‑Service (NaaS) strategy, where programmable APIs expose network capabilities as on‑demand services. Quality‑on‑Demand (QoD) APIs act as the intent interface in this model, enabling applications to request session‑level performance characteristics. QoD requires sophisticated new network control‑plane orchestration to address heterogeneous enforcement, multi‑access behavior, and session continuity issues across Wi‑Fi, private 5G, and public 5G domains.

QoD refers specifically to the CAMARA Project’s implementation, which provides standardized APIs for dynamic multi‑access orchestration, session‑level QoS enforcement, and integration with 3GPP control‑plane functions, including PCF, SMF, NEF, and NWDAF together with UE‑side ATSSS for traffic steering, switching, and splitting.

In this session, we will overview the strategic importance of QoD APIs, the global scale of the emerging NaaS domain, and detail Open Source technologies that are foundational to the industry.

OpenBao is an OpenSSF project and a fork of HashiCorp Vault. It is an open-source secrets manager with support for static and dynamic secrets including identities and certificates.

The OpenBao community recently landed support for horizontal scalability, formerly a Vault Enterprise exclusive feature, in partnership between multiple organizations in the community.

This session will outline the design and development process of the feature, showcase how horizontal scalability improves performance on Kubernetes, and highlight future improvements the community is considering to offer write scalability.

If a production incident hits on your first day, can you debug it? Or if you are a senior engineer, do you find it impossible to download your years of debugging intuition into a new hire’s head?
Kubernetes troubleshooting often depends on undocumented decision paths: where to look first, which signals to trust, and how to turn a sea of logs into a testable hypothesis.

In this talk, we introduce KUBE-RCA, an open-source incident assistant that plugs into your preferred external LLM and provides real-time cluster evidence (metrics, logs, events) as structured context. Using a ReAct loop, the agent proposes hypotheses, runs an allowlisted set of read-only commands/queries, ties each claim back to evidence, and publishes a concise RCA draft directly to your team channel.

We’ll share the design decisions behind our guardrailed execution loop and how we encode SRE intuition into prompts and checks. You’ll walk away with an understanding of how to make incident response more systematic. So engineers of any tenure can resolve issues faster, with less senior interruption.

Kubernetes Dynamic Resource Allocation is now being adopted by hardware vendors implementing drivers for accelerators and high-speed networking. But as real-world AI workloads hit the spec—disaggregated inference across multiple nodes, topology-aware co-location of compute and network devices, per-workload hardware configuration—gaps are emerging. This Birds of a Feather session invites platform engineers, infrastructure teams, and Kubernetes contributors to compare notes on where DRA works well and where the spec leaves teams filling in the blanks themselves. We will use the AWS Trainium Neuron DRA driver as a concrete reference point to ground the discussion, then open the floor: What attributes should be standardized across vendors? How should multi-device atomic allocation be handled? What does the community need to build so that DRA delivers on its promise across all accelerator hardware?

Is it DNS or is it network? & why is answer always DNS? In K8s, most clusters quietly rely on /etc/resolv.conf on every node for all non-service name resolution. At scale, this dependency becomes a liability, reducing caching & observability while causing fragile forwarding, higher latency, upstream resolver overload, and host-level divergence in multi-cloud environments

This talk will explain how K8s actually consumes node DNS today & then present two solutions: CoreDNS for large, complex clusters; dnsmasq for medium scale that bring deterministic caching, controlled forwarding, traffic steering, and real visibility—turning DNS from black box into an engineered system component

Attendees will gain a framework to choose DNS implementation strategies by understanding trade-offs based on cluster scale, workload type and platform maturity. We’ll compare CoreDNS & dnsmasq, surface real failure modes, & show how different designs affect latency, reliability, & blast radius.

When our platform served a SaaS company growing from
9M$ to $100M ARR, we faced a decision every scaling team encounters: stay on Heroku or migrate to Kubernetes. We chose migration. This talk shares the real lessons from moving 47 microservices to EKS while maintaining 99.9% uptime.

I will cover the migration patterns that worked, the ones that failed spectacularly, and the operational changes that made the difference between success and rollback. You will learn how we reduced API latency from 700ms to 70ms, why our first three migration attempts failed, and what "production ready" actually means when your customers include enterprise clients with strict SLAs.

This is not a vendor pitch or theoretical framework. It is a practitioner's account of what happens when you bet your platform on Kubernetes and have to deliver.

Attendees will leave with a realistic migration checklist, common failure patterns to avoid, and honest metrics on what to expect during and after migration.