<!channel> how is everyone doing? We're working on a brand new feature and here's a sneak peek of what we're cooking up. We're building out infrastructure monitoring for your resources and this is a sneak peek of the 2 views ⎯

list

and

map

. Looking to understand more on how we can shape this to make infra monitoring even easier for you. And what are some of the issues you face with other monitoring tools that we can do better.

Looks mad, I love the topological aspect behind how this visualisation works. Any chance one can modify an existing SigNoz cluster to get a preview of this locally?

Any chance one can modify an existing SigNoz cluster to get a preview of this locally?

Not right now, we're still looking to flesh out the direction of how we want this feature to shape up before shipping anything. But happy to hear more thoughts on use cases or functionality that we might be missing right now.

Hey the challenge with that type of design is extensibility in future. I'd suggest you think more of an approach where you can have sections and allow people to pick them. or map their existing data to your defined entity types somehow. i.e. a host is a host is a host and that's fine, but things like containers, are they docker ones? ecs ones? fargate ones? eks ones? etc. having a tab dedicated to k8s seems redundant when you can vertically group that stuff and make it flow from left to right, especially now how have modal's for overlays of data. One of my largest clients using signoz uses a lot of ECS for example - if we have ecs containers, where do they fit in your UI? Having a mode blade'd approach and a entity groups -> entity selector, will make it easier to extend imo

more of an approach where you can have sections and allow people to pick them. or map their existing data to your defined entity types somehow.

Fair point. We have been having a lot of internal discussions around allowing users to map their construct with how we're defining things to prevent extra setup and configuration headaches. As of now, that is in a pretty nascent stage, but this would probably need some time to get properly fleshed out.

if we have ecs containers, where do they fit in your UI?

Fair too. We also have been discussing on how to plug these logical entities in. The current construct very much just works with hosts and kubernetes, but the separate consideration where one entity/term can be a sub-category for another umbrella (like docker containers or ECS, as you mentioned). I think the primary challenge is defining these logical groups/entities with minimal conflict and that's something we need to work on. But if you do have more thoughts on how we can logically group this, I'm all ears.

The approach Newrelic takes is quite interesting to this - have you seen this type of thing before? https://github.com/newrelic/entity-definitions - they have their open standard for metric types etc that generate entities in their platform. So as long as someone sends data in their right shape an entity is born. The beauty of that as well is you can then generate dashboards on the fly for your common (or golden as they call it) metrics for that entity type.

Yes, came across this while researching on the mapping 😅 This is pretty interesting. We were considering this as part of a config-driven workspace setup where you just upload a file and your workspace gets tailored to exactly how you need, instead of just relying on the GUI.

yep exactly - one size doesn't necessarily fit all. It's always a bit of a balance to try and find sane defaults, but still provide complexity for those that need it in the areas where they need it. A configuration based approach to allow for opt in of the complexity is a good one, especially if that's automatable too via your APIs.

@Darren Smith how strongly do you feel about the lack of extensibility? And as someone who has set up SigNoz for clients, I'd love to hear more thoughts what some of the most obvious pain points you think are to set things up. And how this gitops-led paradigm would help in scenarios like setting up and getting more extensibility?

Hey @Shuvam Manna - For transparency - I just have 1 client using signoz - AkzoNobel (they do paint stuff and own things like Dulux) - for their China region. It runs observability for their Magento based B2B Platform. We had to customise things from scratch as they needed ECS deployment pattern, so we took signoz and made it fit. Our ops and dev teams are now using it daily for Traces/Logs/Metrics and understanding what's happening to their platform as they've not been able to see anything properly before. Most of my experience over the last 10 years has been New Relic of which I have dozens of clients on. The Flexiblity I tend to use elsewhere is more of a config based "terraform-ops" style patterns when it comes to configuration management of the platform - as at a larger scale everything from being able to quickly provision "types of alerts" to common dashboards to users to permissions, etc.