kentoh - Fotolia
Edge computing technology can ease bandwidth pressure points
Stockpiling data at endpoints makes sense in a growing number of applications. With storage at the edge, concerns about storage bandwidth and data costs can be mitigated.
Every day, connected devices used in consumer, commercial and industrial sectors generate massive amounts of data. Yet, communicating all of this information to a data center or the cloud is often very expensive, due to limited bandwidth and an exponential increase in the storage of potentially stale data. By using edge computing technology, one can selectively filter useful data from the overall noise.
"The edge is where the physical world meets the digital world," observed Lee Caswell, vice president of storage and availability products for VMware. "With the proliferation of endpoint inputs and the limitations of bandwidth, it's clear that we will need distributed processing and persistent, shared storage that allows value to be extracted far from the core data center or cloud."
According to Ramya Ravichandar, director of product management for edge intelligence software provider FogHorn Systems, the capabilities of edge storage make it particularly helpful in use cases where the costs of sending the data become a concern. "Companies can push this data to the cloud [or data center] at optimum time points," she noted.
Top use cases
Edge computing technology is often used when an organization has knowledge workers scattered across office locations, job sites, manufacturing facilities and other remote facilities. "With users needing to work effectively in their silos or across sites collaboratively, having files and data sets [on-site] makes their access and ability to modify much faster as opposed to pulling files from another location -- sometimes across the globe," said Shirish Phatak, CEO of distributed storage solutions provider Talon Storage.
For many years, video surveillance and "black boxes," such as flight data recorders, ranked as the top edge storage use cases. "Today, with the advent of the internet of things [IoT], the use cases for edge storage have become almost endless," said Todd Loeppke, lead CTO architect for disaster recovery at business continuity firm Sungard Availability Services. "In general, edge storage is typically utilized in devices that have inconsistent or limited network connectivity." Another common use case is in situations such as self-driving vehicles communicating with each other, where real-time decision-making is critical.
Edge computing technology is also playing a critical role in the evolution of storefront retailing. Many top clothing brands, for instance, are installing custom digital displays, smart mirrors and smart racks in their brick-and-mortar stores, transforming the shopping experience. "You can add an outfit to your fitting room right from the screen -- just go back and try it on," said Wayne Carter, chief mobile architect at data platform provider Couchbase. Such technologies demand 100% availability and guaranteed millisecond response times, regardless of network disruptions. With edge data storage, those are met.
Meanwhile, in healthcare, there's a growing movement toward tracking pharmaceutical usage in patients, automatically recording time and dosage information on mobile phones. "This information is communicated back to the caregiver, enabling them to make more informed decisions and resulting in significantly better quality of care and outcomes," Carter said.
A growing number of organizations are also relying on edge computing technology to speed IoT-driven data analytics. "Edge storage, when combined with real-time streaming analytics, is very powerful," Ravichandar stated. "This combination drives intelligent data collection and storage, rather than just storing all data that comes in."
Yet, for all of its power and potential, storage on the edge remains unsuitable for some tasks. "Edge storage is not very useful for continuous interaction with central servers when you need to continuously interact with centralized storage," explained Gaurav Yadav, founding engineer of distributed storage solution provider Hedvig. "For instance, financial institutions cannot provide a stock exchange with edge storage, since they need to synchronize with their central location for every transaction." This type of storage also isn't suitable for preserving large volumes of data, since the technology isn't generally scalable.
Avoid common mistakes
Ramya RavichandarDirector of product management, FogHorn Systems
The biggest mistake novice edge data storage adopters make is failing to identify the technology's fundamental purpose. "Organizations must first understand why they need edge storage," Ravichandar said. She advised managers to ask themselves several questions, such as: What are the business goals that will be achieved through the deployment? Is it the raw data that is of value, or are there operational insights that also need to be recorded? Are there compliance regulations that force this type of data storage, and if so, how can those be addressed in a less brute-force manner?
Another mistake frequently made when planning to deploy storage at the edge is underestimating the amount of storage that will be needed in the future. "With the advances in artificial intelligence and machine learning, data that may have been irrelevant at the time of the decision might end up giving you a competitive edge or saving you money in the future," Loeppke observed. "You don't want to go too small, as the location and quantity of devices makes upgrading storage space difficult."
Additionally, some edge computing technology adopters have tried to minimize cost by selecting comparably inexpensive disk-based media over flash storage. "This is a mistake, because it does not factor in the reliability benefits of flash for remote deployment," Caswell said. "Successful organizations are selecting flash for the best long-term total cost of ownership."
Jeff Fochtman, vice president of global marketing at Seagate, noted that it's also important for organizations to architect a data lifecycle strategy in order to control data at the point of origin, extract value and manage long-term costs. "You need to look end to end," he advised. "Basically, it's not about looking at the edge, the cloud or security independently; think about your holistic data strategy, and see where edge solves a need -- not the other way around."