Elnur - stock.adobe.com

Vast Data: Big shifts promised for an AI future

Purveyor of Universal Storage says AI and GPU-based compute will dominate, and is set to benefit with its (relatively) cheap, deep and quick QLC-plus-Optane offer

Vast Data will triple engineering investment and promises “very ambitious products” as the company orients towards a future in which machine learning and artificial intelligence (AI) will dominate IT and investment around it.

That’s according to chief marketing officer Jeff Denworth, who spoke to Computer Weekly this week. He set out a vision of Vast as a rapidly rising star with a product that fits a future of very large volumes of data – generated from machine and human activity – from which organisations will want to quickly gain insight.

Vast Data sells what it calls Universal Storage, based on bulk, relatively cheap and rapidly accessible QLC flash with Optane (or near-equivalent) fast cache to smooth input/output. It is file storage, mostly suited to unstructured or semi-structured data, and Vast envisages it as large pools of datacentre storage, an alternative to the cloud.

Despite the tie to specific hardware, Vast sells only software – it’s based on a containerised control plane – and with customers able to monitor and control fleets of Vast deployments anywhere across the globe via Uplink Cloud Management.

Vast is in a momentous period in which company revenues have outstripped all expectations and business performance is at record levels for a storage supplier of its size and age.

In late July, for example, Vast announced deals worth $30m with US Federal customers. Quarterly revenue increases, when they have revealed them, have been in the hundreds of percent, and Denworth is also keen to talk up Vast’s performance relative to investment secured – and in contrast to – other storage companies, singling out AWS and Pure Storage.

“You don’t get many tech companies that grow at a rate of 3x or 4x annually without burning through cash,” said Denworth, citing Pure and AWS as examples of those to have done that. “Investors are salivating over the growth we’ve shown combined with the efficiency over the long-term.” Vast, he said, has been far less dependent on funding rounds, and when it has tapped investors it has had specific goals in mind.

Vast uncontested?

How has Vast achieved this? He puts it down to the company having little competition in terms of technology, while also benefitting from relatively few big ticket sales rather than many small deals.

“Unlike other big tech adoption events, we’re uncontested in universal storage,” said Denworth.

“Companies are killing themselves to get heard, while we stand on an island – of flash at the price of hard drives – and we target people that spend big,” he said. “We figure we can spend the same amount of effort or less closing a few big deals than we would on lots of small ones.

“We have half the number of people Pure Storage do to achieve the equivalent goal and that means spending far less. We do deals with people who cut big cheques.”

He probably had in mind some of Vast’s big US federal customers, who contributed $20m-plus to the company’s sales earlier this year.

Denworth also claimed Vast is performing superbly in terms of net recurring revenue, with customers going from, “a few hundred thousand to over $20m” in terms of spend with the company.

A 3x expansion in engineering

But what does all this mean for the future of Vast Data? Denworth said the company would “start tripling the size of the engineering team, as long as the business allows it”, and promised a revised and strategic “full vision” looking out five or 10 years – “Elon Musk-style” – but not until 2022.

However, when asked about the technological fruits of such engineering investment, he remained tight-lipped. That’s intriguing, because such a strategic drive surely signifies a major effort to re-orient the company’s offer.

While refusing to elaborate, he did, however, set out the wider context that Vast aims to play in. Namely, a world where computing is characterised by the pervasive use of machine learning.

“When you look at AI, we’re still in the earliest phases of an event that’s about to happen,” he said, citing research that says revenue generated around AI will surpass even that of the rise of the internet.

The reason for that? “As computing gets closer to understanding real-world events, the amount of processing power will increase exponentially,” said Denworth.

That means, he said, that while customers now are largely concerned with preserving data, “in future, it’ll be all about flash and extremely high-powered processing, with flash that doesn’t care how many times you go back to it”.

Machine learning is the future

“In the next 20 years we’ll see a new class of application,” said Denworth. “It won’t be about transactional data, it won’t be about digital transformation. What we’ll see is computing coming to humans; seeing, hearing and analysing that natural data.”

According to Denworth, that will require a new computing framework, and see “very ambitious products” announced by Vast.

He went very quiet at this point in terms of detail, but did give out what might be described as clues.

“We are conscious of not computing in one datacentre, and using unstructured and structured data,” said Denworth. “We are also conscious that data has gravity, but so also does compute when you get to the high end.”

Read more about Vast Data

Vast is a company based on container-based software, so in theory that allows for a great deal of portability. But at the same time, Vast is very much linked to specific hardware configurations.

“We can drop pre-configured hardware into a datacentre in two-and-a-half hours,” said Denworth.

That all points to farms of high-performance processing backed by Vast’s rapid-access bulk storage, and linked by something like its Uplink Cloud Management.

So, no big gaps in terms of what Vast has already in its stack. Maybe there’s something more. It’s a case of “watch this space”.

Read more on Software-defined storage

Data Center
Data Management