The hyper-connected Philipe Nicolas is the editor of the Storage Newsletter. Each year he asks the leaders in the industry to weigh in with their predictions on what will transpire in the coming year. The list of companies is long and the respondents are distinguished. So long in fact that, Phillipe has to break it into two lists (A-M, N-Z). Covering more than 20 pages and with 46 respondents - there is a tremendous amount to be gleaned. The key, however, is to take a step back and consider it on the aggregate level.
The reason you have to step back should be clear. Each of the panelists has an angle, a lens on which they view the world, resulting in predictions that, in one way or another, favor their particular organization. Still, these are smart people and what they are predicting is genuine, even if it is slanted in their favor. No one wants to sully their reputation by making a prediction they know to be wrong - just because it supports their company.
If you read all 23 pages and you code each prediction (generally three per executive) you start to get a consensus picture. Granted the coding is subjective, but I would be comfortable sharing my work with anyone who is interested. What follows is an aggregate view of the predictions along with some commentary and honorable mentions.
First, the esteemed group sees a big year for AI/ML and Edge. The panel named these areas 30% of the time - well ahead of the other themes.
While AI/ML should not be a particularly big surprise, it is, afterall, the defining technology trend of the past decade, its role in storage is interesting and of note. AI/ML was mostly mentioned for its ability to extract value from massive data sets, to determine what data belonged where (advanced tiering) and to drive automation in IT. The AI/ML theme often mentioned object storage, specifically for its ability to handle unstructured data, log data, etc. but also for its ability to scale to the size needed to optimize AI/ML’s power.
The other clear leader was the edge. While the edge is certainly a hot topic, the fact that it was tied with AI/ML was notable. Driving the prediction is the growth of data on sensors, IOT, 5G etc. There seemed to be a strong bias toward storing and computing that data at the edge as well - something that is tied to solutions that are software defined, lightweight, cloud-native and ultimately S3 compatible. The storage medium for the edge will be NVMe according to almost every prognosticator. This is a function of falling prices for Flash which was mentioned by 10% of the panel.
Following these two trends were the prediction of more hybrid and more multi-cloud. Hybrid cloud was slightly ahead in terms of responses (22%) and if you added Private Cloud into the mix would actually be the #1 mentioned prediction. They are, however, not the same thing so we kept them separate. The message is clear, however, the panel felt strongly that data would continue to reside on premises as well as in the public cloud and that the private cloud would actually grow faster than the public cloud.
Multi-cloud (multiple public clouds and private cloud) also drew attention with the common theme being how it is managed (ineffectively at present with much needed improvements on the way). The ability to have a single pane of glass with superior security to mitigate risk and comply with privacy regulations were at the top of the improvements list. None of the panelists felt that the shortcomings of management would slow the growth trajectory. It is worth noting as well, that none of the panel saw the hybrid/multi cloud game as a zero sum endeavor. They expect all parties to continue to grow as data proliferates.
Next up was Kubernetes. We would have anticipated seeing this topic higher on the list, but it is fifth and probably reflects that fact that this is a storage crowd, not a DevOps crowd. Still, the panel of prognosticators clearly recognized that Kubernetes was going to have a major impact in 2020. The commentary focused mostly on the fact that it was still difficult and that it needed to get easier while recognizing that the Kubernetes train had not only left the station but was picking up speed.
The list tails off a bit after these top five. There are another five areas that pulled five mentions apiece (11%) which include object storage, flash pricing, software defined storage, private cloud and cyber. Most of these are mentioned in conjunction with the previous areas so I will not go into further detail.
There were some interesting takes that we thought were worth mentioning.
One was the rise of tape. This was a bit of a headscratcher. Granted both of the predictions came from tape vendors, but still it violates our thinking around making a prediction that would be wrong just because it aligned with your business. We are not suggesting tape goes away, but it is probably a reach to suggest it is poised for a breakout.
The other was the argument for simplicity. Several different prognosticators made an articulate and persuasive case for the value associated with complexity reduction. We couldn’t agree more with this sentiment. Simple doesn’t have to mean feature-poor. Simple is, in fact, harder to build than complex. The storage world needs more simple - particularly as we look forward to AI/ML and Kubernetes.
Lastly was the lack of takes on file. Only two folks predicted a big year for file and they did so tepidly at best.
Again, the full prediction list is worthy of your time.
As always we welcome your comments. It will be interesting to look back and assess the wisdom of the crowd at the end of the year.