Virtualization Technology News and Information
Excelero 2019 Predictions: Latency is the Final Frontier

Industry executives and experts share their predictions for 2019.  Read them in this 11th annual series exclusive.

Contributed by Tom Leyden, vice president of corporate marketing, Excelero

Latency is the Final Frontier

The world of data center technologies and particularly storage is experiencing fundamental shifts, a few tremendously overhyped technologies, and some transformative innovations that may not catch your attention immediately. Most of us are skeptical about self-proclaimed "visionaries" and their predictions - me included.  Yet what we're witnessing is so powerful that we'll likely look back to 2019 as a pivotal year when data centers became vastly more efficient, scalable, and readily managed.

Here's our team's storage take on winners and also-rans for 2019:

1.  Latency is the final frontier, which NVMe can and will overcome in 2019.

IOPs don't matter anymore, given a single off-the-shelf flash drive can deliver 4-5 million IOPs/drive. Latency for scale-out architectures has become the final storage frontier: these same flash drives offer under 200 microseconds of latency but the challenge is to deploy NVMe at scale while preserving those low-latency characteristics. Expect this challenge to be solved in 2019.

2.  IoT applications will drive storage needs.

We have a long way to go for AI to become what the marketing folks claim it is  -  machines actually learning things autonomously. Some claim IoT will need AI to make any sense of all the data connected devices gather, but I think this is just another way of feeding the AI hype. What IoT needs is strong analytics capabilities - low-latency storage, powerful GPU's and applications.

That said, there is no way to ignore IoT: billions of connected devices are capturing unprecedented amounts of data that can give us valuable new insights. This data needs to be stored and will define new storage requirements: new levels of scale and predictable performance (latency!).

3.  GPU computing will require scale-out tier-0 storage.

Whether it is for high-performance analytics, or actual AI and machine learning, GPU's are very much on the rise. Deploying GPU's at scale for analytics-based applications requires scale-out tier-0 storage to feed the processors with larger data sets, versus using local flash only.

4.  SDS will give integrators a leading role.

Integrators will play an increasingly important role in bringing to market specialized storage solutions for cutting edge applications like AI/ML, VR/AR, and other fast-growing sectors. They have access to scalable, high-performance software-defined solutions, a wide selection of HW components and top-notch applications such as data analytics etc.

5.  More use cases for flash and NVMe.

Continuing the 2018 trend of NVMe becoming a commodity, we expect enterprise NVMe and Enterprise SATA SSD to achieve price parity on a Euro/GB basis in 2019. Further, while not crossing the mythical Euro/GB cost line of spinning drives, advancements in density such as 128TB SSDs vs. spinning 3.5" drives at 14-16TB force end-users to consider SSD not just for performance, but also for scale-out use-cases. The advantages in density, power (and hence cooling) and random access ability of these SSDs will make people start to doubt the viability of spinning media vs. the convenience, far greater speed access and reliability of solid state media.

6.  NVMe over TCP/IP.

2019 is the year for NVMe over Fabrics via TCP/IP. The combination of a rich server ecosystem with all sever vendors offering NVMe platforms, and the ubiquitous nature of Ethernet and TCP/IP, will lower the hurdles to trial and adoption - driving a revolution in next generation SAN.


About the Author


Tom Leyden, VP Corporate Marketing at Excelero, is  an accomplished Marketing Executive who has excelled at creating thought leadership and driving growth for half a dozen successful storage and cloud computing ventures. Tom's track record of technology startups includes data deduplication pioneer Datacenter Technologies (acquired by Symantec), early cloud innovator Q-layer (Sun/Oracle), Amplidata (HGST) and Scality. He was also responsible for productizing DDN's Web Object Scaler (WOS). Tom has strong relations with influential press and analysts and is often asked as a speaker and content contributor to educate technology audiences on upcoming trends and paradigms.

Published Monday, January 21, 2019 7:32 AM by David Marshall
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<January 2019>