Last year at this time I made three, what I considered bold, predictions. I didn’t want, and still don’t, to be obvious in my predictions; I wanted to make it more difficult and try to forecast something outside of the norm. Last year, I predicted that the percentage of organizations that had pulled at least one workload back from the cloud would increase. I predicted that NVME over Fabrics adoption would increase with FC growing at a faster rate. And finally, I projected that many AI projects would fail due to lack of content awareness in tools and technologies.1
In a 2019 study of IT storage decision makers, 55% of organizations pulled a workload back from the cloud.2 I am going to count that as something I got right. For anyone who has not read my earlier writing on workload repatriation: These moves are not an indictment of the cloud. The moves, though common, only happen with a handful of applications, and IT organizations that pull workloads back generally continue to use the cloud and think highly of public cloud services. These moves are often the result of overuse of a “lift and shift” approach to shifting workloads to the cloud, not conducting proper due diligence, and not refactoring the apps. These events will likely eventually fade overtime as cloud adoption and expertise grow, but until that time, these efforts increase the cost of the cloud adoption.
On NVMe over fabrics, it is still early, but the results so far are different than I expected. Based on current research data, the preference for NVMe over fabrics adoption is trending toward Ethernet.3 The prevalence of Fibre Channel combined with how easy it is to upgrade an existing FC environment to NVMe over Fibre Channel still suggests that NVMe over FC will have strong adoption moving forward, but the jury is still out on this.
On AI projects failing due to lack of content awareness, I might have been too bold on that claim. The value created by AI initiatives is so high that inefficient data management adds cost but is not causing the initiative to be considered a failure. I still expect, however, that understanding content will be a major storage focus area moving forward.
1. See ESG Blog, 2019 Data Storage Predictions: More Cloud Missteps, FC Is Back, and Finding Data Holds Back AI, January 2019.
2. Source: ESG Master Survey Results, 2019 Data Storage Trends, to be published November 2019.
3. ibid.