Machine data collection and analytics can scale.
Over the past decade, we have learned how to scale machine data collection and analytics. More importantly, we know how to avoid building failed architectures and implementations.
There is no one-size-fits-all, but practice for building architectural foundations does exist, and we do it better than most.
Machine data ownership.
Something little known is that when you buy a machine data analytics platform, you never truly own your data. During your vendor selection process, ask this simple question:
If I moved to a different data collection and analytics platform, would I have to recreate connectivity to my machines and assets?
You can expect over 90% of vendors to respond with a resounding, “Yes, you will need to recreate machine connectivity.”
The practice to avoid.
Complete vendor lock-in. A change in a data analytics platform should have zero effect on upstream machine connectivity.
Machine data interpretation.
Machine data needs interpretation… somewhere. The closer to the data source you can interpret and standardize what your asset is telling you, the better off you will be.
The reasons are simple:
Subject matter expertise exists near the asset, not once you move machine data bit-for-bit into an analytics platform.
Raw machine data and dozens of Fieldbus protocols create data management chaos downstream.
Interpret once, not within every application consuming the data.
The practice to avoid.
“Stretching” machine data without interpretation into downstream applications.
Machine data standardization.
Machine data interpretation leads to standard representation and shared understanding across software applications and business roles.
When you and I discuss a machine controller's execution state, we agree that it is either actively executing instructions or ready to execute instructions.
MTConnect is one of the standards for removing this guesswork.
The practice to avoid.
Inventing your own standard to interpret machine data to information.
Visibility.
Following the basic tenets of data ownership, interpretation, and standardization makes providing essential visibility much simpler and quicker to achieve.
We have open-sourced several pieces of software demonstrating industry best practices.
The practice to avoid.
Building applications that do not scale across a variety of machines and technologies.
Still have questions?
Contact us to learn more about the work we do.