After spending years debugging distributed systems across microservices, queues, APIs, and cloud platforms, I kept running into the same problem:
We have logs.
We have metrics.
We have traces.
We have error tracking.
But we still struggle to answer a simple question:
“What actually happened?”
Most observability tools treat these as separate concerns:
Metrics live in one system
Logs in another
Errors somewhere else
Audit trails are usually an afterthought
So I started building something different.
Over the past while, I’ve been building Provenance — a structured observability platform focused on causal tracing and auditability across distributed systems.
The core idea is simple:
Everything is an interaction.
A user action, API request, service event, background job, or error becomes part of a structured, linked model.
That means:
Errors are tied to full request chains
Metrics have context
Audit trails become queryable
Root cause analysis becomes dramatically easier
Some things Provenance currently supports:
Structured interaction tracking
Distributed tracing via Unit of Work (UOW)
Hierarchical span tracking
Metrics + alerting
Error tracking with fingerprinting & issue grouping
Event-driven subscriptions/actions
Dashboards + widgets
Webhooks / Slack / adapters
Compliance-friendly audit history
One of the bigger design goals was:
👉 not just collecting telemetry
👉 but structuring it into something actionable
I recently put together a short deck explaining the philosophy and architecture behind it:
http://stdiolabs.dev/assets/provenance_active_observability.pdf
Would genuinely love feedback from:
DevOps engineers
SREs
SaaS founders
Platform teams
Anyone dealing with distributed systems at scale
Still early, still evolving — but I think there’s an opportunity for a different approach to observability here.
You make a strong point here. Modern observability feels fragmented because every layer of the stack has its own dashboards, alerts, logs, traces, and workflows. Teams often end up switching between multiple tools just to understand one issue. A lot of engineers now say the problem is no longer “lack of data” — it’s lack of connected context across systems.
What’s interesting is that as systems become more distributed, observability itself is becoming another distributed problem. Microservices, cloud infra, APIs, AI workloads, and third-party integrations all create visibility gaps that traditional monitoring was never designed for.
I think the winning platforms in the next few years will be the ones that simplify correlation instead of just collecting more telemetry. Engineers don’t need more dashboards — they need faster answers and better context. Great discussion topic.
This comment was deleted 2 months ago