Observability must start pre-production to prevent failure and ensure trust.
In the automotive and manufacturing world, system complexity is not just a function of software—it’s tightly intertwined with hardware, compliance mandates, and decades-old legacy systems. I’ve seen how critical it is to embed observability as early as possible in the software lifecycle.
Too often, organizations treat observability as a production concern. The reality is, by the time you’re putting out fires in production, the damage—operational, reputational, even regulatory—is already done. That’s why my mantra is simple: shift observability left.
You can’t afford to wait for production to find problems. In regulated, engineering driven businesses, preproduction is where observability must lead.
This means designing an observability strategy that starts in pre-production. In a manufacturing environment, where software increasingly drives core operations, connected vehicle platforms, and IoT-linked diagnostics, the margin for error is razor-thin. Whether it’s a performance issue, an integration failure, or a downstream latency choke, catching it early is non-negotiable.
One area that remains a challenge is working with fragmented tooling—tools that monitor specific stacks or layers without the context to tie events back to business impact. Root cause analysis suffers when signals are scattered across multiple systems — only to find the issue stemmed from a legacy hardcoded delay, like a 30-second thread sleep which may take weeks to diagnose. That kind of inefficiency is what modern observability can eliminate, if applied correctly
Another real-world pressure is justifying ROI for observability investments. Whether you’re on-prem, hybrid, or SaaS-first, costs add up fast when licensing, infrastructure, and skilled resource needs are factored in. The key is understanding how observability reduces unplanned downtime, accelerates root cause resolution, and supports compliance and risk mitigation.
Let’s also not forget cybersecurity and compliance. As observability platforms evolve, they must interlink with security telemetry—especially in a post-DPDP, zero-trust world.
One of my biggest lessons? Observability isn’t about monitoring more things—it’s about guessing less. The less time people spend chasing symptoms, the more they focus on outcomes. And when you apply that mindset early in the lifecycle, you don’t just prevent incidents—you build confidence.
In high-performance, compliance-bound environments like ours, experience begins not at the front end, but at the far left of your pipeline. And that’s exactly where your observability must begin too.
–Authored by Dr Pawan Kumar Sharma, Chief Information Security Officer, Tata Motors
