The explosion in the amount of data generated by iot devices is forcing companies to rethink their iot storage strategies.Dilip Sarangan, Frost&Sullivan's global research director for Internet of things and digital transformation, says, "right now, a lot of companies don't really know what data is really useful, but once analyzed, it can help them make business decisions."Gartner, a research firm, estimates that 24 billion internet-of-things devices will be generating huge amounts of data by 2020.So the question is whether to keep it, and if so, where to store it.
Analysts at Gartner say cloud computing storage, which has been a major target for the past few years, is now moving toward internally deployed storage and edge storage.And that decision is based on specific applications and specific locations."Many organizations that have the Internet of things don't need to go to the cloud.They mainly want to be internally deployed and localized for processing and storage.The problem of delay, in particular, will be hard to solve even in the next decade.It makes no sense to transfer data to the cloud when most applications on the edge need at least real-time analysis.
EricBurgener, IDC's vice President of infrastructure systems, platforms and technology research, said decisions about where data analysis takes place will drive the purchase of capacity to determine whether data is stored in the cloud, in a live data center, or on the edge.Burgener said that layered iot storage options in the cloud (for example, from hot to cold storage) are less relevant in iot applications because data can be stored locally.For tiered storage, he says, the basic concept is that all data doesn't have to be stored in the same high-performance, high-cost storage area.
"The enterprise just needs to be there to provide the data that must be delivered at high performance and to move to a lower cost data layer, which is data that the enterprise does not need to be on-site or frequently accessed," he said.Use cases that require immediate analysis of data, such as self-driving cars, typically do not use cloud storage.Some time-intensive use cases allow data to be sent back to a central location where machine learning is applied to gain insight.Certain industries and applications generate large amounts of data, including utilities, building automation systems (lighting, heating), and monitoring systems.For example, surveillance cameras transmit about two megabytes of data per second
"This combination of iot storage types is suitable for many enterprises," Burgener said.You might see hundreds of megabytes and tens of megabytes of storage in a data center, but less than 20 megabytes at the edge."Edge devices don't necessarily need a lot of processing power or storage, just capture the data and send it back to a more centralized location for processing.Depending on their location, these devices need to be rugged and have enough power to sustain the data until the batteries can be replaced.
"There is a really broad continuum of device types, so they have to develop different types of storage devices to run at very low power in iot devices.This is often the new solid-state storage technology."For example, data that self-driving cars use to decide whether to park will be processed at the edges to make an immediate decision, but once the decision is made, the data is discarded, Burgener said.However, decisions about whether the vehicle needs to be refueled based on mileage are collected at the edges and sent back to the remote location for processing and storage
"There is no unified storage solution for the Internet of things," said NatalyaYezhkova, vice President of infrastructure systems, platforms and technology research at IDC.Everything is driven by a specific use case.Some of the larger workloads may be sent to the cloud for analysis and then returned to the internal data center for fine-tuning."You also need to consider how comfortable the enterprise is with cloud storage, how much data is transferred, and how much control it has over bandwidth when it needs to move a lot of data
Structured data collected by iot devices provides information about machine characteristics that has been transmitted to the data center for analysis.Surveillance video produces large amounts of unstructured data that may need to be analyzed in real time or frequently, so it is not reused.In addition, the amount of edge storage is determined by several factors, including the distance between the edge and the core.For example, low-power sensors on offshore oil RIGS may require additional storage and battery power, depending on how often data is uploaded from the device.
"If the main goal of the Internet of things is to be achieved, data can't just sit idle, it needs to be analyzed to improve operations and customer experience," she said.Most iot data is unstructured and easy to store in a public cloud.All of the major cloud providers offer scalable storage systems for virtually no data entry fees.The cloud also provides big data analysis tools for large jobs in local data centers
When considering storage options, enterprises should ensure that data management is aligned with the size of workloads and applications."There is no set formula for when to use the cloud and data center," Yezhkova said. "for some companies, it requires repeated verification to get the final result.