Customers still find IoT to be hard. It's still a bit nebulous, they don't know where to get started.
Our goal at Oracle is to make IoT easy and to focus on business outcomes.
The suite of SaaS solutions, first introduced earlier this year, is designed to be configurable using a declarative approach, with no software coding required. Oracle says its customers have said this makes the solutions two to three times faster to deploy than more custom, platform-based alternatives. There are four applications in the suite, covering asset monitoring, production monitoring, fleet monitoring and connected worker (which focuses on safety).
All the applications run on an underlying platform-as-a-service in the Oracle IoT Cloud. Today's announcement adds three new sets of capabilities, as well as new partner integrations, into that platform. These significantly deepen the functionality available in the application suite.
AI and machine learning
Predefined Artifical intelligence (AI) and machine learning (ML) functionality is now embedded in the applications, designed to be configurable without the need for specialist data scientists, according to Oracle. For example, there are built-in operational analytics that can parse machine data to detect anomalies, predict failures, and prescribe corrective actions. The analytics can also provide intelligence to feed back into performance improvements.
Specialized time series algorithms support automatic model creation and tuning to analyze IoT data. Predictive models combine IoT data with business data from a range of enterprise applications including manufacturing, maintenance, service, logistics, warehouse, financial and ERP. This ability to merge data from IoT and business sources is one of Oracle's special strengths, says Lionel Chocron, Vice President, Oracle Industry IoT Solutions:
Because we have a very deep understanding at the application level, we have this ability to bring the insight from the IoT application into the enterprise application to provide a business outcome.
An important new capability is the ability to create a digital twin to represent a corresponding physical asset. This uses an object model to capture the operational and behavioral dimensions of the asset and supports multiple views into current, historical, and predictive data.
The digital doppelganger can be used to run simulations and 'what-if' scenarios to model the effect of changes in environmental conditions or business processes, as well as aiding in preventative maintenance. This helps reduce capital expenditure and speed time to deployment of new solutions, as well as minimizing downtime and optimizing asset performance, says Oracle.
Digital Thread for SCM
One of the most unique capabilities is Digital Thread, which provides an end-to-end view of the entire manufacturing lifecycle. It uses IoT and integrations into supply chain management applications to connect traditionally siloed elements in real-time throughout the digital supply chain.
Templated automated workflows allow a manufacturer to track items from product design and order fulfillment, to manufacturing and product life cycle management, to warehousing and transportation, through to logistics and procurement. As well as providing better visibility, this can also enable new business models, such as dynamic demand planning or a move to XaaS. Nainani says:
Now you can track an asset all the way. This is enabling a very responsive supply chain because we have automated all these workflows using IoT.
Expanded partner ecosystem
Oracle has tripled the number of partners it has certified for its IoT cloud in the past six months, ranging from device makers to systems integrators. In addition, today it unveils a collaboration with Mitsubishi Electric to develop an IoT platform for smart manufacturing, based on Mitsubishi's FA-IT Open Platform for factory automation, which has been developed with Oracle Cloud.
It's refreshing to see a vendor grounding its IoT strategy in real-world business outcomes, rather than the breathless hype often heard in this space. During its pre-briefing yesterday with analysts, Oracle cited several customer examples. Noble Plastics has realized immediate savings from implementing IoT-enabled automation. Vinci added sensor-driven building automation and discovered it was able to introduce new services to customers based on IoT data. SoftBank has used geolocation sensors to track usage of electric scooters by tourists on a Japanese island, for billing and compliance purposes.
But much of the rhetoric surrounding today's announcement looks ahead to broader use cases than these very specific examples. At a larger level, the issues around data normalization, security and integration that have bedeviled other IoT platforms will loom large for Oracle's customers too, as diginomica contributor Brian Sommer notes in his observations immediately after yesterday's briefing:
- It’s the beginning of the game for ORCL and IoT. In baseball terms, they’re in the early innings. Lots of early goodness here but winning is still a ways off IMHO.
- As such, the technology will permit firms to wire up key parts of the plants and some aspects of the value chain/supply chain.
- But, any discussion re IoT is going to trigger a slew of related matters involving big data, algorithms, etcetera. To that end, what we still need to understand is:
- How well will ORCL’s solutions accommodate huge volumes of non-anomalous sensor data? Will they offer a pre-processor to summarize or discard large quantities of non-interesting sensor log data?
- How open will ORCL’s pre-provided algorithms be to customers and business partners? The best algorithms are not black boxes. The best algorithms have feedback mechanisms that customers — not vendors or partners — can adjust and enhance as the customer tries to make the algorithms smarter and the data more accurate.
- One of the biggest problems customers have with digitizing their firms is that they don’t have one ERP system. They often have several, from different vendors with different data models. Even if a customer has standardized on ORCL, they may be running many different instances of the software with different data meanings in place. This makes it tough to do an enterprise IoT environment if the base data isn’t all the same. How will ORCL solve this?
- How will these IoT data feeds and their companion algorithms be supplemented with non-ORCL and non-customer datasets?
- Many times, the speakers mentioned how quick this solution is to deploy but no word on the ongoing costs and manpower needed to keep the solution adding value to the customer. Will customers need an army of quants, data scientists, etcetera to keep this afloat?