As we continue to rethink storage and its changing role in the industry, it is also valuable to examine the changing trends in the industry to predict how the role and use of storage will need to change in response to the industry. Storage system architecture and design cannot remain static in the face of changes surrounding it, rather, it must adapt to meet the new use cases and traffic patterns.
One such emerging trend which will dramatically impact storage architecture is the rise of the “Internet of Things”, or IoT. What is the IoT? Simply put, the IoT refers to a world where all manner of everyday items have a capability to monitor themselves and to communicate their status via external networks (see Gartner definition). What kinds of items would do this?
a. Smart devices. We have seen already a growing level of function and intelligence in consumer technology in the form of tablets, phones, e-readers, and the like, with connected watches and glasses not far away. These devices achieve their maximum value when internet connected and serve as a two-way, personalized portal for the user.
b. Expensive devices. John Deere is known for the connectivity technology built into their tractors. Cars today are highly computerized and can monitor themselves for impending faults, with detailed information available for technicians and, increasingly, for the drivers themselves. Given the price tag associated with these purchases and the long term value of the data gathered, the initial price of the sensors and connectivity technology is trivial, which will lead to a vastly increased set of data gathered and available to drive innovation and improvements.
c. Dumb devices. This is the category which will see the biggest growth and which will drive the biggest changes in the IoT world. RFID and other passive sensors can be embedded in clothes, shoes, retail items, and more. While each such sensor is limited in its inherent capabilities and will likely have no inbuilt connectivity, other classes of smart devices can sample these sensors and upload the data, thus dramatically expanding the set of devices for which data is available and dramatically expanding the usefulness of the data.
d. Collecting devices. As dumb devices proliferate, there is a need to sample the data from the dumb devices for later processing and/or analysis. Collecting devices may themselves be embedded in cars or smart devices, or they may be standalone components, such as elements of a warehouse tracking automation engine.
All Politics Are Local
In the end, the major impact of the IoT will be on end users. Already we are seeing the changes of the IoT, with smart devices becoming a collection and control point for surrounding devices, e.g. television remotes replaced by downloadable phone/tablet apps, home thermostats connected and remotely controlled, mobile airplane boarding passes automatically updated in response to flight delays or itinerary changes, exercise apps which track activity at a detailed level and allow both historical or peer-sourced comparisons comparisons, and more.
Now imagine the possibilities when the data can be cross-correlated across sources! I have an EZ pass transponder in my car, which allows automated toll payments for me. Now imagine that when I get on the highway, the point of entry, direction of travel, and day/time are noted – the automated system looks at my history, calculates the most likely exit point(s) for me, and sends me with an ETA or traffic update, assuming that the drive conditions differ from normal. Or imagine that I use a mobile boarding pass, my e-reader is notified of the projected flight time, and it determines that I will finish my current book before the flight finishes, so it consults with a central database, finds a few books that I may wish to read and auto-downloads them, making them available for purchase during the flight (even if I am disconnected at the moment.)
What Lies Beneath the Tip of the Iceberg
The benefits of the IoT are only possible with the right infrastructure beneath.
a. Sensors everywhere. This is already occurring, and will only accelerate, as the technology becomes cheaper and more integrated into manufacturing and other processes.
b. Decentralized collection of data. Tablets and phones are starting to play this role, serving as local access points, with internet connectivity to central engines with a larger data repository and greater capabilities.
c. Centralized storage and processing of data. This will be the biggest effect on the storage industry, as the reams of data collected are stored, analyzed, and transformed into useful formats. Data will be analyzed historically to detect patterns, across users to detect trends. The processing of the data at this scale will be far beyond the capabilties of the edge phone/tablet devices, although the edge devices will be the primary channel for access to this information.
d. Object storage will serve as the primary long-term repository for data and will serve as the bulk upload target for newly generated data. Analytics may be run locally, or file and block based systems may be utilized where a higher degree of compute and I/O performance is needed. Overall orchestration engines will arise which will treat the entire data center, and all its assets, as a single environment, with data, compute, and network integrated to provide an optimal data processing experience.
e. Data security. Within the central, integrated data repository, security controls must be in place to validate incoming requests, ensure that data visibility is appropriately restricted, ensure that system operators and administrators cannot inappropriately access data. Data security, encryption, anonymization, and threat detection technologies will all play a key role in this infrastructure.