Cost Implications of IIoT Analytics

Industries rely on analyzing sensor and process data to prevent unexpected breakdowns and optimize processes. Traditionally, companies have sought the cloud to perform such data analytics tasks. The cloud is viewed as a central aggregator for both data and computation. This requires many companies to move sensitive data to the cloud and pay high compute costs.

Cost implications of IIoT Analytics are not immediately apparent

For example, a multi-site manufacturing enterprise can pay up to $30 million annually for cloud storage and compute. This does not include the cost of highly skilled DevOps and software engineers needed to work alongside data scientists to keep the analytics pipeline functioning smoothly. It’s no surprise that we hear executives complain about costs being “astronomical” and that its “literally killing us”.

What we do know, however, is that costs are only expected to increase as more companies mature through their journey of digital transformation and their data analytic needs grow. In fact, some companies are beginning to search for alternative solutions such as Edge computing, which comes with its own challenges.

Economics of a real world IIoT Analytics case study

Vibration Analytics for a major manufacturing company was way more than anticipated

To put things in perspective, we present a real-world case study on vibration analysis for a large manufacturing company. The numbers presented herein are based on our own experiences with industrial analytics. The analytic task at hand was to build a pipeline for processing data from 20,000 vibration sensors spread across 50 manufacturing sites.

First, let us consider the effort incurred in developing the entire pipeline by an internal team (internal to the company). Their objective was to build a cloud-based pipeline for vibration monitoring and fault diagnostics. Here is a breakdown of the cost components.

• At a monitoring cost of $4000/plant/month, the estimated annual cloud cost alone amounted to $2.4 million annually.
• The pipeline required 8–11 full time employees (4–6 IT staff, 2–3 data scientists, and 2 engineers) to support the cloud analytics workflow costing approximately $2.3 million annually in labour costs.
• The total internal expenditure was close to $4.7 million annually.

Now let us consider the cost of using a third party cloud-based analytics provider offering the same services.

• An aggressive, competitive price point ranged anywhere between $10,000–$15,000/plant/month. Based on our past experience with cloud-based analytics providers, we validated the contract price for a similar data science project to be approximately $7.2 million annually.
• Accounting for a bulk discount, amortization and good negotiation the cloud analytics contract for 50 sites could cost roughly $4–6 million annually.
• However, the time to fully integrate the local data collection and migration to the cloud took anywhere between 6–8 months with some providers insisting that data be transferred to their cloud account.

Bear in mind that these numbers are for just one single analytics application, i.e., monitoring and fault diagnostics. In reality, there are multiple such projects that a company would be interested in pursuing (e.g. prognostics, process optimization, etc.) but may simply not have the budget for.

Blockalytics : Leveraging decentralization to reduce IIoT analytics costs

Blockalytics is driven by decentralization at its core which immediately leads to a significant reduction in labor requirement. We estimate this reduction to be close to 50% to 75% since data scientists can themselves architect the entire analytics workflow through our automated decentralized computational platform. Moreover, since computation is now decentralized across site-level Edge computers/devices, we estimate a reduction in cloud costs by at least an order of magnitude. Using Blockalytics would imply leveraging the cloud mainly for storage and archival purposes only. Therefore, the bulk of cloud costs would mostly be for storage and archival. As a result, Blockalytics is strategically positioned to provide the same cloud-based analytic capabilities at a fraction of the costs that companies incur today.

IIoT Analytics for all: Blockalytics for SMEs and OEMs

Moreover, due to the lower cost of ownership, Blockalytics is an attractive choice to a wide variety of sectors such as SMEs (Small-to-Medium Enterprises) and OEMs (Original Equipment Manufacturers). Many OEMs struggle to offer an analytics platform to their customers because of the high barrier of entry discussed above. In fact, based on our discussions with customers, only 10% of OEM providers have an analytics platform that offers value to their customers leading to a market potential of 90% in the OEM domain.

Further, due to decentralized sharing of insights, Blockalytics delivers an unbeatable turn around time in order to start realizing a healthy ROI. Currently, it could take anywhere between a few months to a year for a cloud analytics solution to generate savings and validate the customer’s ROI. The reasons for such high setup times could range from a diverse set of data collection tools used at each site, with unique data collection protocols to legal, organizational, and security/privacy hurdles to transfer data to the cloud. These challenges are compounded when multiple (potentially competing) companies are involved, say an OEM and its pool of customers.

By design, the Blockalytics platform naturally addresses many of these challenges, lowers cost and significantly reduces the time-to-deployment (and hence ROI) from several months to a few days.