Four Smart Decisions for Building a Digital Manufacturing Platform

Building a Digital Manufacturing Platform

Table of Contents

To see a Google/Sight Machine webinar on how manufacturers are starting their digital transformation journey go here.

By: Jenn Bennett, Google Office of the CTO;  Nate Oostendorp, Sight Machine CTO

Manufacturers embarking on digital transformation projects face a series of early decisions that will ultimately determine whether the projects succeed in transforming the company or become failed science experiments.

Whether they realize it or not, these decisions will determine whether the project accomplishes its central objectives of building a digital manufacturing platform. Whether it addresses a single problem or can be used to drive ongoing change; whether it is cost-effective to maintain and keep secure; and whether it will break down data silos rather than create new ones.

The decisions include whether to build a one-off solution or a reusable data model; whether to embrace the security and power of the cloud or build only local solutions; and whether to build or buy.

Through our work helping some of the world’s largest manufacturers achieve transformational results, we’ve learned quite a bit about what works and what doesn’t. Here are four smart strategies for success.

1) Don’t Just Build a Smart Factory … Build a Smart Enterprise

Many initial digital manufacturing projects are one-off data science projects or do-it-yourself applications built by data scientists for site-specific problems in production or quality. These aren’t scalable, aren’t cost-effective and often fail to achieve adoption in even the original target factory.

Projects to analyze data from one machine or one plant at a time will often be unable to deliver the big wins companies are seeking through digital transformation. Analyzing data from multiple machines — including identical machines, classes of machines or similar machines from different vendors — makes it far more likely that the key variables will be identified. To transform the enterprise, manufacturers need to deploy common solutions across multiple factories.

Software written for a single piece of machinery or factory is unlikely to have the architectural flexibility and scalability to be extended across the enterprise. It is unlikely to be usable with other machine types or be capable of taking a broader view as new variables of interest emerge. A company that invests in a set of scattered local projects is likely to be left with orphaned applications.

2) Don’t Build When You Can Buy

Most manufacturers relying on their own custom-developed software will find it difficult or impossible to keep up with efforts by competitors who are using the best-of-breed software that is increasingly being developed primarily for cloud usage.

Many manufacturers don’t initially realize how much time, effort, skill and money are required to build custom manufacturing analytics solutions and manage them on-premises. For all but the companies with the largest IT budgets, it will be far too expensive and inefficient to maintain all the in-house teams that can compete with the very best specialist providers of commercially available software tools.

Every line of code written is a liability. That code needs to be documented, maintained, updated as external systems change, and kept secure as vulnerabilities emerge in legacy software.

Companies assembling their digital manufacturing platforms using commercial and open source components will get their systems up and running far faster than those building from scratch. These components tend to use open interfaces that will let manufacturers swap them out later with alternate modules if better options emerge or their needs change.

3) The Cloud is Protected by the World’s Best IT Security Talent – Use It!

The major cloud platforms deliver far better security than any traditional on-premises solutions can achieve. They employ the world’s best IT security talent. In a relentless arms race, they spare no effort or expense to stay ahead in the lead, deploying far more resources to security than the largest manufacturers could afford.

Ongoing security is much easier to manage on the cloud, where deployment control allows for consistent security policies. Instead of having many point solutions, security updates in the cloud can be rolled out within hours of a vulnerability — often by the cloud providers.

Just-in-time manufacturing depends on constant communications between sales, manufacturing and suppliers. While this communication may often happen on a private network, the data is still transmitted from one location to another and exposed to security risks at every location. Few if any manufacturers can provide security across a private network comparable to the centralized, uniform and state-of-the-art security the major cloud providers enforce.

4) Consider a Hybrid Cloud-edge Architecture … It’s Not One or the Other

The right combination of cloud, private network and edge computing resources will vary depending on corporate policies, the type of factory equipment used and the specific needs of the projects under development.

A cloud architecture is critical for enabling the benefits of enterprise wide visibility into manufacturing data.  Centralized data enables supply chain wide optimization, cross plant benchmarking, and end-to-end product traceability. For these capabilities, we have found that for most customers a very lightweight edge solution is appropriate, for getting data to the cloud. Often, it will be most efficient to use an edge device to perform initial processing of the data and then send selected fields up to the cloud, after compressing and encrypting it.

There are a number of edge solutions that focus on low latency use cases involving localized monitoring and control of individual machine or lines.  While this can be a great way to begin building digital manufacturing capabilities, scaling capabilities across the enterprise will eventually require a hybrid cloud-edge approach. Defining a hybrid architecture that can support both the short and long term capabilities will ensure the initiative can serve as the basis for true enterprise-wide digital transformation.

To see a Google/Sight Machine webinar on how manufacturers are starting their digital transformation journey go here.

Time Is Money Manufacturing analytics can provide a real-time view into your operations using the data that’s already being generated throughout your factory, giving you a head start on identifying and solving problems that might arise. Download Now

Picture of Nate Oostendorp

Nate Oostendorp

Co-Founder and CTO at Sight Machine Nathan co-founded, worked 9 years as an architect for, and has developed several other successful online communities. Nate has also worked in industrial controls. He holds an MS in Information Science from the University of Michigan and a BS in Computer Science from Hope College and he has contributed to Forbes.

Curious about how we can help? 
Schedule a chat about your data and transformation needs.