User Interface (UI)

[pleshakov]

I wonder if we should add UI to NGF.

For example, a UI can visualize:

• The topology -- how many data plane replicas are there.
• The configuration on a data plane.
• The graph of resources -- HTTPRoutes, Gateways, Services, Secrets. For example, it will be possible to see which HTTPRoutes attached to which listener.
• Once we support Policies, also Polices --- which Policies apply to which resources
• A real time visualization of control plane (not data plane) metrics -- something similar to https://demo.nginx.com/ but for control plane metrics.

With visualizations like above, the UI can make NGF easier to operate and troubleshoot for cluster operators.

[mpstefan]

Maintaining a UI is a big commitment - I would expect all engineers would participate, not just a dedicated "UI person."

Fortunately, I think we can provide a lot of what you're suggesting above through something like Grafana. I see things like topological maps, control plane metrics, even the graph of resources as an observability feature (since it helps you understand what components are attached to where).

Perhaps when we look add adding more first class observability features we could provide additional endpoints with the data we'd need to build dashboards that give exactly what you're looking for above.

[pleshakov]

Perhaps when we look add adding more first class observability features we could provide additional endpoints with the data we'd need to build dashboards that give exactly what you're looking for above.

Could you provide examples of such data? I can only think about metrics stored in a data store (Prometheus), not real time data.

[f5yacobucci]

I'd like to meet users where they are at and not provide visualization in isolation. Creating something new that only shows Gateway config and state relative to NGF concerns will force users to look at multiple pages to understand their app, e.g., if users are already using Apache Skywalking will this help or require cross-referencing between the systems?

[pleshakov]

I'd like to meet users where they are at and not provide visualization in isolation. Creating something new that only shows Gateway config and state relative to NGF concerns will force users to look at multiple pages to understand their app, e.g., if users are already using Apache Skywalking will this help or require cross-referencing between the systems?

that makes sense!

I'm thinking more along the lines about exposing the internal state for debugging/troubleshooting, specific to a particular instance -- so when you really need to understand that instance. You can visualize such state using UI (it will look nicer), but at the same time, that's not as important as being able to dump that information.

Higher order systems can import that data to build additional intelligence with cross-referencing.

[brianehlert]

If N+ support is added, the nginx.conf can be defined in such a way that the objected are reflected in the NGINX Plus dashboard.
NIC has been following this pattern already. And uses zones (for example) to annotate hostname and upstream service name in the dashboard to make local troubleshooting easier.
Also, advanced use cases like cache will simply 'show up'.