OpsMill Presents at Networking Field Day 41

OpsMill Infrastructure Automation and AlOps, Built on Data You Can Trust

Watch on YouTube
Watch on Vimeo

Damien Garros of OpsMill presented the company’s vision for simplifying infrastructure automation through Infrahub, its platform for capturing infrastructure intent. The session covered the evolution from imperative to declarative and GitOps automation, a three-layer model of business, technical, and component intent, native version control, and the schema, transformations, and generators that let organizations build custom automation on a common foundation. The talk opened OpsMill’s appearance at Networking Field Day 41 and set up a multi-domain orchestration demo from Alex Gittings.

Garros described OpsMill as a company of about 35 people, founded almost four years ago, that builds Infrahub and recently took over the Nornir project. Using the Network Automation Forum’s framework, he placed Infrahub in the intent block and Nornir in orchestration and execution. He argued that network automation remains stuck between imperative and declarative approaches because existing tools can’t capture desired state well. In his model, the business layer describes services and designs, the technical layer holds IPs, VLANs, devices, and interfaces, and the component layer contains device configurations. Today these live in separate places: NetBox or Nautobot, Ansible, and custom tooling. Infrahub unifies all three with native branching and version control. Its extensible schema supports transformations that render configurations for any vendor, and generators that turn business requests into technical data. Answering delegate questions, Garros said existing networks can be imported through tools like IP Fabric or Slurp’it, while greenfield customers generate full configurations before devices arrive on site.

Garros called build-versus-buy a false choice, since every successful organization in this space has ended up building its own business layer, often creating a maintenance bottleneck. Infrahub aims to supply the foundational 80 percent (data access, permissions, and version control) so teams build only the 20 percent unique to their business. By his account, this cuts projects from years to months or even weeks. He said AI alone doesn’t solve this hard problem, but it sharply reduces the effort of building on Infrahub. OpsMill is designing for a future where agents both run infrastructure from trusted intent data and help build the platform itself. For understaffed enterprises, he pointed to professional services, partners, and AI-assisted development as ways to get started.

Personnel: Damien Garros

From User Request to Multi-Domain Orchestration with OpsMill Infrahub

Watch on YouTube
Watch on Vimeo

Alex Gittings of OpsMill demonstrated how Infrahub turns a simple user request into coordinated changes across networking, security, and compute. Using a simulated ISP and data center environment, the session showed self-service requests through a Backstage portal, generators that build technical data from business intent, branch-based review, and deployment through Nornir and Kubernetes. Gittings, a solution architect, presented at Networking Field Day 41, with Damien Garros joining to answer delegate questions.

The Containerlab environment combined an Arista spine-leaf VXLAN fabric, a Nokia SR Linux WAN, a branch site, a Juniper SRX firewall, and a Kubernetes cluster. Gittings noted that teams can start from common schemas on the OpsMill Marketplace and change them safely in isolated branches. The core schema, including IP addresses, can be extended, and users set their own uniqueness rules, such as per VRF or per site. Playing a user named Alice, he requested an internal wiki through the portal. Generators then assigned a prefix, created firewall policies, and produced DNS, Kubernetes, and fabric configurations. The request showed up as an Infrahub branch with 131 changes, credited to Alice rather than the portal. An admin reviewed the resulting firewall, Kubernetes, and AVD-generated Arista configurations before merging. Infrahub can manage IP allocation within a delegated supernet and sync results back to Infoblox or BlueCat.

Delegate questions focused on how Infrahub fits with existing systems. Gittings said the portal is interchangeable, and Backstage reconciles requests made outside the portal. When asked about removing configuration, the presenters said safe deletion depends on declarative methods such as commit replace or vendor Terraform modules, which work out what needs to be removed. Garros said OpsMill deliberately avoids direct integration with any single vendor. The demo used Nornir for the switches and firewall and a separate system for Kubernetes, and some customers run Infrahub on top of Cisco ACI. He described vendor platforms as vertical, per-domain solutions, and positioned Infrahub as a horizontal business-intent layer that sits above them and complements them.

Personnel: Alex Gittings

Behind the Demo – OpsMill Infrahub + Arista AVD

Watch on YouTube
Watch on Vimeo

Damien Garros, co-founder and CEO of OpsMill, explained the Arista AVD integration behind the earlier Infrahub demo. The session covered how OpsMill embedded Arista’s open-source PyAVD library in Infrahub’s transforms and generators, replacing AVD’s YAML files with data managed in Infrahub to produce validated Arista configurations within a broader multi-domain orchestration workflow. Garros presented the segment at Networking Field Day 41, where OpsMill announced the integration publicly for the first time.

Garros described AVD as a popular open-source Arista project used in hundreds of environments. It follows the same three-layer intent model as Infrahub: users define a design at a high level, such as a layer 2 or layer 3 spine-leaf or a service provider network, and AVD generates the technical layer and device configurations. It also produces test plans and per-device documentation, which Garros said customers especially value. AVD is normally Python wrapped in Ansible and integrated with CloudVision. OpsMill’s version uses no Ansible at all. Instead, OpsMill built Infrahub schemas that map to AVD’s data model, so Infrahub serves as the state machine for every layer. Arista’s logic runs inside Infrahub’s transforms and generators, and Infrahub still provides version control, the API, and the UI. Because Arista’s library generates the configurations, users don’t have to write their own Jinja templates. Garros said this means AVD covers about 15 percent of the 20 percent of custom work Infrahub leaves to customers, leaving only around 5 percent.

Garros clarified that AVD configures only the Arista devices. Because the firewall and compute data live in the same connected platform, though, a single change can drive the fabric through AVD alongside firewall and Kubernetes updates. Asked about other vendors’ validated designs, he said OpsMill is happy to work with any vendor and already has an online example based on Cisco validated designs. He added that customers integrate Infrahub with ACI, NDFC, Fortinet, F5, and even optical networks. Garros noted that AVD’s advantage is that Arista stands behind its configurations, but the same approach can be rebuilt for any vendor on Infrahub.

Personnel: Damien Garros

OpsMill Intent: AI Ops You Can Trust

Watch on YouTube
Watch on Vimeo

Xavier Martins-Rivas, field CTO of OpsMill, explained how Infrahub’s intent model provides a trusted foundation for AI operations. The session covered moving institutional knowledge out of individual engineers and into structured, versioned, validated data, a top-down knowledge graph linking services to configurations and devices, and guardrails that let AI agents act on infrastructure safely. Martins-Rivas presented the segment at Networking Field Day 41, noting that Infrahub now has more than 600 live deployments.

Martins-Rivas argued that most tools reflect their builders’ assumptions, so organizations fill the gaps with “Bob,” the long-tenured engineer who knows why a given switch must never be touched. That works until Bob goes on vacation. Intent replaces that tribal knowledge with structured data, shifting from teams carrying complexity to teams managing trust. He contrasted observability-style knowledge graphs, which build upward from devices and struggle to map impact onto custom business services, with Infrahub’s approach. Infrahub starts at the service layer and works down to configurations and devices, so users can trace relationships in both directions. He described four pillars of trusted infrastructure: a schema that models everything, versioning to move forward and back, validation against the organization’s own requirements, and generators and transforms that produce configurations and artifacts. When a delegate asked about test frameworks, he said validation rules are entirely user-defined to match each use case.

For agents, Martins-Rivas described two modes. Existing skills let agents build new services, such as an L3VPN, while MCP servers answer questions like what depends on a given device. To guard against hallucination, every agent change goes through a branch and a validation process that can plug into existing workflows like Jira tickets. He compared this to laying train tracks with gates so agents can’t go off the rails. Critically, Infrahub, not the agent, holds the memory. Its immutable database records every change and who made it, enabling rollback, and agents can be swapped or retired without losing anything.

Personnel: Xavier Martins-Rivas