Ready to boost your edge computing environment's storage? Find out some best practices to follow, including how to integrate with multi-cloud environments and NVMe-oF.
Storage at the edge used to be about ensuring remote offices and transient staff had adequate ways to capture and retain the data they generated. Today, IoT devices have changed all that, generating huge amounts of data while often having limited connectivity. Getting that data to centralized storage can be a slow and overwhelming process.
Enter edge computing and storage, where processing and storing data happens close to where the data is generated.
Edge computing storage is all about the methods and technologies used to collect and retain information at the network periphery. The goal is to keep data as close to the originating source as possible and, in some cases, do initial processing there.
Options for data storage at the edge include micro-data centers and high-performance fog computing servers installed at remote locations and used to replicate cloud services.
Storing and processing data closer to where it originates improves performance and saves time and money over sending it to a central data center or the cloud. The following five tips provide insight into how best to approach and implement storage at the network's edge.
Storage options at the edge
There are several ways to store data at or close to the network edge. Some IoT devices come with a small amount of built-in storage. In addition, vendors add SD cards to provide extra storage on the device or elsewhere at the edge. Micron Technology's microSD cards, for instance, can be used to add storage to video cameras to provide backup to network storage. Micron's SD cards can support three years' worth of continuous video recording, and they provide a mean time to failure rate of 2 million hours, according to the company.
Devices with large internal DAS are also available for storage on IoT implementations that include edge servers or server clusters. Whatever you use, industry expert and IT journalist Alan Earls recommends rethinking data management and storage when implementing an edge computing environment.
Applications where edge storage works
Edge computing storage is particularly useful for devices that have inconsistent or limited network connectivity, such as sensors used on airplanes or in agricultural applications in rural locations. It's also critical for devices, such as self-driving vehicles, that take in a lot of data and engage in real-time communication with other devices and vehicles.
Other uses include storefront retailing with custom digital displays, smart mirrors and smart racks; healthcare monitoring; and any IoT application that requires intelligent data collection and storage, using real-time analytics. Veteran IT journalist John Edwards examines several applications that require the availability and millisecond response times that edge storage provides.
Storing and processing data closer to where it originates improves performance and saves time and money over sending it to a central data center or the cloud.
Know your data's value
Organizations embarking on edge technology often fail to clarify upfront why they're using the technology. The edge storage approach taken will depend on knowing what it is you want to achieve and how the data collected is important to that goal. Is it the raw data that's of value or the insights gleaned from the data after it's analyzed?
Another frequent mistake is low-balling the amount of storage that will be required in the future. Rapidly changing analytics and AI technologies can turn today's irrelevant data into valuable information in the future. Designing enough storage today for tomorrow's data needs is easier than upgrading in the future. Edwards investigates these and other mistakes to avoid with edge storage.
Understand edge computing storage in a multi-cloud world
Edge computing combined with a multi-cloud storage environment can get complicated. Success depends on extensive preparation that covers analysis of how data is collected, transferred and retained. Consultant and writer Robert Sheldon looks at questions to answer before attempting to integrate edge and multi-cloud, including:
What type of data will be generated and collected?
Where will the data be generated?
What data will need to be retained and for how long?
What data will be transferred to the cloud and how often?
How long will that data be retained in the cloud?
How will data be managed?
How will security, compliance and disaster recovery be handled?
Where NVMe/TCP fits in
The NVMe over TCP transport binding spec enables the creation of storage systems designed specifically for the edge and edge cloud infrastructures. NVMe SSDs provide the consistent and low-latency capabilities that edge computing requires. NVMe-oF, which connects hosts to storage across the network fabric, makes it possible to scale storage and compute resources independently with little impact to latency. And NVMe/TCP provides this flexibility without requiring specialized switches and network interface cards.
Edwards breaks down the advantages and disadvantages of using NVMe/TCP for edge computing storage. He also delves into how to get started and what's next.
