Traditional data historians place today’s manufacturers at a distinct disadvantage. Built for individual plants, these legacy systems leave operational data fragmented across facilities, equipment, and proprietary platforms. As manufacturers add more sensors, production lines, and connected assets, the number of unique data streams they manage is exploding. Historians built around fixed tag counts weren't designed to scale with that complexity, and that's exactly where blind spots emerge.
"These legacy tools have become a bottleneck because the shape of industrial data, and how we work with it, has changed. Ingest volumes are higher, context across systems is key, and the interfaces stay closed to the tools that need the data," explains Benjamin Corbett, Solution Engineer for InfluxData. "Manufacturers end up unable to consolidate operational data across plants, integrate it with analytics and AI, or extend it into new use cases."
Those limitations show up in three ways: siloed operations, an inability to scale with modern telemetry, and closed architectures that make data difficult to access and extend.
1. Siloed At Every Plant
Manufacturers often struggle with a fragmented tech stack. Their data is notoriously siloed between multiple facilities, different OEMs, and separate systems. While data historians work well within individual plants, manufacturers increasingly need visibility across these systems and facilities to support centralized operations, analytics, and AI. When you don’t have a unified interface, correlating this information becomes nearly impossible.
"Every plant has its own collection of systems. You've got ERP managing inventory, DCS running the process, MES coordinating production, and machine telemetry capturing what's happening on the factory floor. Each serves a purpose, but they rarely work together. What starts as disconnected data within a single plant quickly becomes an enterprise-wide challenge across multiple facilities," Corbett explains.
For example, tracing a batch number back through to inventory stock should be a simple use case. A company might want to create a report about the effectiveness of a particular batch, errors associated with an individual batch, group results by manufacturer ID, or look at incident reports associated with a particular asset type. But without creating custom integrations into the data historian, conducting this level of in-depth analysis would require a sizable team mired in a manual process.
2. Inability to Scale
Manufacturers are capturing telemetry at increasingly finer resolutions, sometimes even down to the nanosecond. For most modern consumption modes (e.g. building ML pipelines), fidelity matters. It's critical to have a solution that retains full data fidelity while keeping pace with that throughput.
"The challenge is that modern use cases demand data volumes that are an order of magnitude higher than legacy historians," says Corbett. "Instead of polling a sensor or a SCADA system every 30 to 60 seconds, now almost always we're dealing with at least second intervals of contextualised signals—it’s a tsunami of data by comparison."
Many legacy historians simply weren't designed for today's telemetry workloads. Measuring 1,000 sensors across 10 machines at sub-second resolution is enough to push many of them to their limits. When the historian can't keep pace with incoming telemetry, neither can the predictive maintenance systems that rely on it. Detecting subtle changes in vibration, temperature, pressure, or current requires complete, high-resolution data, not a platform that's struggling just to ingest it.
Even when manufacturers overcome the technical limitations, they often encounter another obstacle: licensing models that penalize growth. Pay-per-tag pricing means every new sensor increases costs, making expansion progressively more expensive.
“They’ve invested in this incredibly deep data historian, but then essentially get penalized for every additional industrial signal,” says Corbett. "This becomes especially painful when you have a high turnover of equipment, but there's still value in retaining the data."
3. Locked Behind Closed Interfaces
Another complication is that most historians are closed by design. Proprietary formats and vendor-controlled interfaces keep operational data locked inside the plant network. Whether the goal is supporting a centralized data science team, connecting a new analytics platform, or enabling another business system, getting that data out often becomes a project of its own.
"Since an organization's data science team is often situated centrally, the only way they can get their hands on the data is going back into plant networks. Each use case becomes an integration project, continually duplicating effort, creating security and bandwidth concerns because many of these remote OT networks are highly egress only," Corbett explains.
The result is a system organizations depend on but can't easily extend, making every new analytics initiative another integration project.
Modernizing Beyond the Historian
Modernizing a data historian starts with building a business case around operational outcomes rather than technology replacement. If the effort is treated as a one-off infrastructure project, it's likely to stall and remain in pilot purgatory.
The next step is to implement a purpose-built time series platform with industry-standard protocols that can run at the edge and consolidate data centrally. Historian modernization enables manufacturers to capitalize on their data to improve operational efficiency and scale successfully.
Use these questions to highlight the need and benefit of adding a time series platform:
- How much engineering effort does it take to answer questions that span multiple plants or systems?
- Are your historians keeping pace with the volume and frequency of telemetry your operations generate today?
- How long does it take to make operational data available to analytics or AI teams?
"Historian modernization has to be an organizational strategic priority, not one person's pet project. When the value isn't owned and understood at that level, corners get cut, priorities shift, and the effort loses momentum long before it delivers," Corbett emphasizes. "Done properly, it's the opposite - a data foundation you build once and layer repeatable, scalable use cases on top of. That's the model where value compounds as you scale, not cost and effort."
Learn how modern time series platforms eliminate these bottlenecks →
