Streaming analytics: How to be prepared for analytics driven IoT architectures?
Next week I will publish my experience using IoT devices as streaming sources for enterprise dashboards on my website. As part of my dashboarding journey, I want to understand how enterprise dashboards perform being fed by real-time and online iot sources. Stay tuned since the results are quite impressive and I was especially surprised by the relative easy set-up of “streaming dashboards”.
The value of enterprise dashboards driven by streaming IoT devices, is for me a no-brainer: I have explained in a previous post how analytics will safeguard 90% of the succes of the Internet of Things. It is crystal clear today that streaming analytics is a necessity for IoT to become a success. And IoT will become a success. The impact is so huge, that streaming analytics – real-time and online – in fact must (!) work properly and rocksolid. There are no excuses. Just to shed a quick light on what I will show shortly from now: no worries; enterprise dashboarding today can perfectly deal with streaming data in real-time.
An easy transition?
For enterprises the migration to IoT applications will be a challenge. As concluded here, this migration will go in phases and enterprises probably first focus on consumer-centric and vertical industry driven IoT applications. In a later stage, they’ll go to what we are really interested in: interoperable IoT applications. Interesting is how they will handle this migration. What challenges do enterprise phase when preparing for the technical architecture to fully leverage from IoT. Let’s have a look.
For sure the transition from a traditional enterprise IT architecture to one that is optimized for IoT applications, will not be an easy one. Elements of companies’ current technology need to be redesigned so they can support billions of interdependent processing events from millions of products, devices, and applications. Because networked devices are always on, companies must be able to react to customer and system requests in real time; agile software development and delivery will therefore become a critical competency. Seamless connectivity will also become a must-have, as will collaboration across IT and business units, which have traditionally been siloed. Moreover, companies must be able to securely and efficiently collect, analyze, and store the data emerging from these refined IT architectures.
Exploring a bit further, we recognize a number of critical area’s that enterprises should focus on when altering the challenges:
Participate in setting industry standards
Explore and rethink modular approaches in application design
Seriously rethink security and data gathering models
Reconsider how to manage existing products alongside new IoT applications
Rethink contracting and SLA agreements for IoT service requirements
Challenges: a closer look
Making the switch to IoT-enabled IT architectures won’t be easy and definately will take time. Applying the above mentioned area’s or principles will help:
Particpate for Industry standards: current IoT applications lack standards for data formats, interfaces or service definitions. Especially if the bigger aim are the interoperable IoT applications, industry standards are required. Early adaports of IoT applications of course have an opportunity to “shape the game”.
Modular application design: with 9 billion connected devices, that will triple (!) the next 10 years, we expect network bandwith issues if we do not think intelligently about the off-loading of data from the devices. Choosing the points for off-loading together with the data processing need tob e carefully done. Probably the device itself will play a role here: preprocessing of sensor data within the IoT device itself, could help. My best guess would bet o embed predictive models into the devices. Predictive analytics allows to isolate those data records that either do not meet the standard pattern (and as such important for analysis) or indicate abnormalities.
Rethink security and delivery models: given IoT device software updates will probably follow a way quicker schedule, than the IoT applications lifecycle, enterprises need to look for a way to enable continuous delivery of software updates. Modular designs will be required so IT engineers can refresh discrete components of an Internet of Things–connected device on a rolling basis without having to upgrade the whole thing. A continuous delivery model requires a kind of dual-speed operating model: one for intervening in customer-application issue and one focussing on the overall system of records application to ensure stability and security. These models require strict integration between IT operations and software-development groups within the company to ensure the speed and fidelity required from Internet of Things applications. The accompanying security models must support this dual-speed operational modus
Refurbish existing products: the newest IoT devices and products may embed technology that is not available in current or older products. However these older generation of products – sometimes even IoT enabled – might be of great use for the development of future products; for instance their data mapping structures are usefull for future products. It might even be necessary to retro-fit older products with IoT afterwards, to benefit from the data and knwoledge for future design.
Cybersecurity: rethink contracting and a lot more. Since every IoT device is an entry point for security issues (hacking, misleading etc), security risks rise exponentially. As such enterprises urgently need to work on digital resilience: embedding methods of protecting critical information in their technology architectures, processes for business-model innovation, and interactions with customers. In parallel they need to win the irreversible trust of the customer; customers must see the value of sharing potentially sensitive information and feel utterly comfortable doing so.