Exponential growth of the Internet of Things (IoT) is rapidly presenting challenges in the areas of security, capacity and analytics that will require the data-center market to restructure how it holds and processes data to effectively handle all that the IoT has in store.
A recent report by the Gartner Inc. research firm argues that the IoT, with its massive quantities of network connections and data, calls for an array of optimally-sized data centers to effectively manage capacity and real-time demands.
“The recent trend to centralize applications to reduce costs and increase security is incompatible with the IoT,” says Joe Skorupa, vice president and distinguished analyst at Gartner.
Within the IoT, all types of network-connected devices and appliances are fitted with sensors that send and receive data. For example, a refrigerator can tell its owners when they are low on mustard, or a thermostat can adjust itself based on whether it senses human presence in a room.
It’s hard to fathom the amount of data these devices will generate. While predictions vary depending on which report you read, Gartner estimates that “the IoT will include 26 billion units installed by 2020, and by that time, IoT product and service suppliers will generate incremental revenue exceeding $300 billion, mostly in services.”
Different data, different centers
Gartner and other experts say that while handling the capacity of data is one thing, making meaningful sense of that data is quite another. All types of data generated by everything from smartphones to vending machines must be processed quickly, usually in real time — a task that can often be performed more efficiently by smaller data centers.
“IoT deals with streaming data from globally distributed sources; transferring these data to a single location for processing will be challenged both from a technology and economics perspective,” says Soumendra Mohanty, vice president of global data and analytics at Mindtree, a global IT services firm. “Rather, there is a greater need to aggregate IoT-generated data in multiple, distributed, smaller data centers where initial processing can occur to provide capture-and-respond kind of values at real time.”
This staged model will help companies mitigate data-storage issues, especially if organizations effectively leverage the devices within the IoT.
“We know that much can be accomplished within devices themselves,” says George K. Thiruvathukal, Ph.D., professor of computer science at Loyola University Chicago. “For example, you can store an extended number of environmental observations on a detection device. Depending on how real-time the application is, you can push data or only certain events of interest upstream, ultimately arriving in a data center. Less critical events can be pushed when the device’s local storage fills up.”
Gartner’s report noted that a decentralized data-center model presents new challenges in areas including management, storage, backup and compliance: “This new architecture will present operations staffs with significant challenges, as they will need to manage the entire environment as a homogeneous entity while being able to monitor and control individual locations.”
To be sure, big data centers are not going away.
“With IoT, big data and BYOD— among a variety of other trends taking off — we’re going to see a different set of data-center requirements emerge,” says Matt Miszewski, senior vice president of sales and marketing at Digital Realty. “These trends will drive the need for both large and small data centers in regions across the world. The size of data centers will be determined by what the data centers do.”
Miszewski described three different kinds of data centers that businesses will need to consider:
- Big data centers: Also known as “motherships,” they will act as repositories of applications and data.
- Distribution data centers: Large regional hubs, they will move data from the motherships down to a retail-oriented level.
- Micro data centers: Comprising one or two racks, they will be located close to population centers so they can receive and transmit data to IoT sensors instantly.
With all of this data moving around all these different places, security is another big concern, according to Loyola’s Thiruvathukal. He recommends companies focus on security and privacy down to the code level, including end-to-end encryption, especially in sensitive environments such as banking and healthcare.
“There is a glaring need for proprietary software to be more open to inspection,” Thiruvathukal says. “Unless you know how it’s coded, can you really trust that it supports security and privacy concerns?”