My predictions for 2018 are very simple: the future will look a lot like the past. We won’t hit a tipping point. Instead, storage will continue to get cheaper, the cloud will continue to be a mixed blessing, and security will continue to be hard.
The public clouds such as Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP) have dominated infrastructure conversation. The changes in the next year will be evolutionary, not revolutionary. Those who've dipped their toe in AWS or Azure will develop a more mature understanding of the strengths, weaknesses, and real costs. Private infrastructure teams are regularly benchmarked against the agility and cost of public cloud. I predict they'll have better tools to help them with decision making such as workload analysis, placement optimization and monitoring.
On the positive side, the cloud will continue to get cheaper. Amazon has lowered the price of S3 storage every two years. I predict that they'll follow suit this coming January, to between 1.7 to 2.0 cents per GB per month in US East. Other public cloud providers, and private cloud infrastructure vendors, will be forced to keep pace.
On the negative side, the cloud will still have highly visible outages. Amazon has suffered major failures of service in 2015, 2016, and 2017. 2018 will be no exception. Some businesses will adopt to this reality and make their applications multi-region or multi-cloud; others will choose to live with the risk.
The other big conversation in technology is security: how do we prevent intrusions and how do we respond to them? I predict in 2018 there will be more large, embarrassing breaches as a result of known software vulnerabilities going unpatched. It almost goes without saying that the responses will be haphazard, not just because of lack of planning, but because it's difficult to admit that you screwed up and caused harm. There will also be new examples of attacks other than traditional network intrusion: disinformation attacks, game-theoretic exploits of smart contracts, and spoofing attacks on artificial intelligence. As technologists, we have an ethical duty to do better, but it's a hard problem and it's not one you can just throw money at to solve.
On the hardware side, NVMe (NVM Express) is stirring a lot of excitement due to its promise of consistent low latency. NVMe drives use PCI Express for data transfer, and have an improved software layer that better uses the capabilities of flash hardware. For direct-attached storage this can lead to fantastic performance numbers. Unfortunately, I predict that many early adopters of NVMe-based storage arrays will be disappointed. Drive latency is often only a small component of end-to-end latency, so even a 10x improvement in SSD latency could be just a 20% improvement in overall latency. Worse, many businesses that erroneously believe they have a storage latency problem will discover that the bottleneck lies elsewhere in their infrastructure stack: poor networking, overloaded hosts, or software that can’t use the full capabilities of a high-performance flash drive.
The growth and challenges of public cloud are the new normal. Reductions in storage costs and innovations in hardware are also expected; it would be only news if they didn’t continue. Security and reliability are old problems. The surprising thing in 2018 will be organizations that continue not to incorporate this reality into their planning.
Unique control with VM-level actions for infrastructure functions including snapshots, replication and QoS make protection and performance certain in production, and accelerate test and development cycles.