Followers of the enterprise storage industry witnessed significant changes in 2018, which can’t always be said about the often stodgy field of how to care for your 0s and 1s. Here are some of the most notable headlines from the year–assuming nothing major happens this month–as we see them from the perspective of covering the industry.
SEE: Quick glossary: Storage (Tech Pro Research)
January: Companies that make data backup software, including Commvault, Dell-EMC, IBM, Veeam, and Veritas, all disclosed that they think your backed-up bits should be put to work rather than just sitting there in case they’re ever needed. Functions such as data mining/analysis and testing new applications for backward compatibility are some examples of ways to get more from loafing backups.
February: Academic researchers presented an idea called corruption-tolerant replication, cleverly spelled “CTRL,” which they said is a better way to recover from damaged clusters of replicated state machines. Experts liked the idea so much that they gave it a best-paper aware at a prestigious conference.
March: Cloud downtime is worse than you realize, Veritas said. A third of 1,200 respondents in a Veritas survey said they expected cloud downtime of 15 minutes per month but actually experienced a half-hour of downtime per month.
April: Nimbus Data explained its 100-terabyte solid-state drive. Its design focuses on reliability, not speed. Two challenges: It’ll be expensive, and it would take 36 hours to write back a full-capacity drive in the event of data recovery from a backup. It could be about five years before drives of such capacity are used in mainstream products.
May: Swordfish is the latest standards method endorsed by the Storage Networking Industry Association (SNIA). Previous efforts focused on hardware, went too long between updates, and weren’t sufficiently scaleable. SNIA management said they learned from those mistakes. Perhaps in 2019 the industry will find out if anyone is willing to give this idea another chance. On paper it sounds great for all storage to conform to a single management standard, but in reality has always been proprietary.
June: What happens if you change the physical shape of flash storage modules? Then you can creatively fit more data in smaller spaces, which is great as long as you can control the heat. Samsung said its drives using the new design allow for 8 terabytes per drive and 576 terabytes per every 2u of rack space.
July: Western Digital closed a conventional hard-drive factory because it’s building a new solid-state storage factory. We’ve known for a couple of years that traditional hard drives are on their way out, so it is interesting to see drive manufacturers transition in tangible ways like this.
August: Intel, following Samsung’s June announcement, said its own new solid-state drive form factor can hold 1 petabyte (1,024 terabytes) in just 1u of rack space. The company did so by lining up memory modules in the shape of a ruler, with one long side serving as a cooling channel, and then putting several rules together in an array.
September: Sometimes one of the hardest things about administering big data is finding the small data you’re seeking. Global file search is still a challenge after all these years, but startups are working on new approaches. For example, Cloudtenna has software to consider the behavior and trends of files and their users, rather than just searching based on Google-like rankings.
October: For all the hype about moving to the cloud, some companies are going the other direction. Reasons could include reduced business needs, less cost savings than expected, too much downtime, too much latency, budget changes, security, a merger/acquisition, different employees with different expertise. All are valid reasons to bring your data in-house, experts observed.
November: IBM bought Red Hat, which immediately made everyone question the future of Ceph and Glustre. Sure enough, the next story was the the Linux Foundation announced a new Ceph Foundation, “able to harness investments from a much broader group to help support the infrastructure needed to continue the success and stability of the Ceph ecosystem”–they didn’t exactly pull punches.
Bonus: All year long, storage companies upgraded their products to better support NVMe (non-volatile memory express), which is a fancy way of saying they’re building servers, storage, and networking gear with internal components specifically designed for solid-state drives. Legacy hardware is designed for conventional hard drives, so while SSDs are an improvement, it’s still like trying to drive your Ferrari on roads made for horse-drawn carriages. NVMe drastically reduces the bottlenecks.