Skip to content

JetStor Makes Storage Predictions for 2019

By Gene Leyzarovich, ACNC/JetStor 2019 is upon us, which means it’s time to take note of trends in the storage industry. Here are five worth considering. NVMe over Fabrics is hot! Nonvolatile Memory Express (NVMe) greatly speeds communications between a host and solid-state storage. It replaces SCSI, which was designed for spinning disks, to take advantage of the greater speeds of SSDs. NVMe over Fabrics (NVME-oF) extends the protocol over Ethernet, Fibre Channel, and InfiniBand networks. NVME-oF delivers lower latency, additional parallel requests, and higher performance across the infrastructure. NVME-oF is a disruptive technology that vendors should be leveraging. Big data analytics is the first and obvious application for NVME-oF, but others that benefit from accelerated storage will follow. And keep an eye on the NVMe Management Interface specification. NVMe-MI will enable the remote management of NVMe devices, enabling them to be discovered, monitored, and updated. The Adoption of Flash-Based Cloud Storage As SSDs continue to drop in their cost per gigabyte, they are increasingly used for cloud storage. SSDs offer faster, lower latency storage than spinning disks, better supporting customers’ needs and applications. They scale easily and non-disruptively, and reliably provide high availability, helping to ensure SLAs are met. Moreover, they reduce space, power, and requirement costs, providing the operational efficiencies that providers covet. With the proliferation of NVMe (see above) further boosting performance, this trend will only continue. AI & Storage: Storage Gets Smarter and, Perhaps, So Do We Artificial intelligence and storage will be more intertwined than ever. AI can help alleviate the stress of managing, monitoring, and maintaining petabyte-sized storage troves. With AI, administrators can identify bottlenecks and optimize where data reside to improve efficiencies and lower costs. But AI demands huge amounts of data. The more data, the better the intelligence, at least in theory, and AI processing itself generates large datasets. As companies turn to AI to uncover insights from data-intensive deployments like IoT, they will need even greater storage. As if routine operations don’t already produce enough data, AI processing will put further pressure on enterprise storage capabilities. Edge Computing is the IoT Engine Edge computing is instrumental in driving the inexorable rise of IoT. By processing data at their source rather than shipping everything to data centers and doing the analytics there, enterprises reduce latency and gain insights in near real-time, which is particularly key in such industries as finance, healthcare, and manufacturing. The ability of edge computing to quickly extract actionable intelligence from vast torrents of remotely-collected data gives IoT much of its value. Expect edge computing to power more IoT applications. SDS & HCI Are Evolving Hyperconverged infrastructure (HCI) offers advantages, but it does have one shortcoming. Compute and storage functionality are tightly integrated and, as a result, cannot scale independently. When you require additional storage, you also have to buy additional computing capabilities, which is not cost-effective. It’s sort of the reverse of traditional silos; rather than have separate pools of resources, you have all resources locked tightly into one pool. As a remedy, vendors are evolving HCI into a “hybrid-converged” model. This approach uses software-defined storage to allow external storage to link to HCI devices. In effect, storage-specific modularity is being added to HCI. You’ll obtain all the benefits of HCI, but will be able to scale storage without paying for everything else.