Having accelerated from start-up to top-five vendor in the red-hot flash array market, Pure Storage is looking for new heights to scale. While it still has plenty of opportunity remaining in the storage segment, it is trying to broaden its horizons with a number of new initiatives, including a data-centric architecture, storage as a service and one of the latest buzzword-bingo catchphrases, artificial intelligence (AI) and machine learning (ML).
“We were ahead of the market in all-flash,” said Pure Storage CEO Charlie Giancarlo in the earnings call earlier this month. “We were ahead of the market with NVMe. And we’re ahead of the market with AI.”
At last week’s PURE//ACCELERATE 2018, its third annual customer/partner event, the company continued its AI push, which first surfaced with the NVIDIA partnership in March. It also made a number of other announcements intended to broaden its reach beyond just faster, smaller, less-complex and more-energy-efficient storage, which has fuelled its meteoric rise, including 40% year-over-year revenue growth last quarter..
“We’re guiding generally to 30+% year-on-year. We aspire to grow just as fast as we possibly can. Part of that is the market, part of that is one’s ability as a public company to scale without wanting to sacrifice quality,” said Giancarlo.
“Last year was a great milestone for the company. We also have $1 billion in the bank. We are cash flow positive and are growing like a bat out of hell,” he added. “We’re not just enterprise storage. We’re in a very great place.”
Pure faces stiff competition in its core business, the c (7%). However, it’s even further behind in the overall enterprise storage market, which rang up sales of $13.6 billion in Q4, compared to AFA’s $1.9 billion. Although it is looking at a total addressable market of $35 billion, the lights are much brighter in the AI segment, which is expected to generate $1.2 trillion in economic value this year, up 70% from 2017, and projected to add just under $4 trillion by 2022.
“The interest in AI by corporations is just off the charts,” said Giancarlo in a recent interview. “At Pure, we are able to…feed GPUs, high speed applications, and AI environments — at the speed they want that data to provide the intelligence companies want to make their businesses better.”
He said since AI is all about crunching huge amounts of data, older, tiered storage systems that rank data by age aren’t nimble enough to grant researchers quick access to even the oldest data sets. “These days, people want access to data, whether it was last week or last year or last decade,” Giancarlo said.
Pure has always had its eye on the future, and this year’s Accelerate announcements continue that focus, said Scott Sinclair, Senior Analyst, Enterprise Strategy Group. “Over its history, Pure Storage has had propensity to deliver technologies that the bulk of the storage industry still considers roadmap items, something to think about in 3 to 5 years. We saw it when Pure was focused entirely on all-flash arrays when the much of the industry was still pushing hybrid. We saw it when Pure announced FlashBlade, which happened at a time when the idea of all-flash file and object storage was thought to be more of an academic argument than an actual practical offering. And at Pure Accelerate this year, we are seeing it again as Pure delivers NVMe technology across their entire portfolio, when many in the industry still see NVMe-based flash as only a premium tier.”
Sinclair breaks DCA down into two aspects: “First, it is a reality of the digital economic transformation we are currently experiencing. Essentially, every company is a data company. Data is more than an enabler for business, it is something that can streamline operations, improve customer relationships, and create new revenue opportunities. If you accept the potential of data, then your infrastructure needs to be designed maximize the value of data, and become data centric.
“Second, Pure is delivering the data centric architecture by extending the benefits of NVMe across the portfolio. Instead of being forced to deploy a separate disparate architecture (e.g. DAS) for the more performant cloud native workloads, companies can leverage a single infrastructure for both traditional workloads and those that demand lower latencies. The question of course is whether they latencies will be low enough to serve the bulk of IT environments, and that will be something to watch over time.
“Pure’s push into AI also extends the narrative of the IT vendor pushing the envelope. You would be hard pressed to find a major IT vendor not discussing AI currently. Yet, many of the solution offerings are still reference architecture based, with a lot of do-it-yourself work involved. Pure’s partnership with Nvidia, helps companies develop their own AI initiatives by streamlining the infrastructure work and allowing companies to focus on the people, the processes, and the data.”
Pure is in a race to either broaden its market, cement its place in storage, and/or to be acquired before it’s explosive growth grinds to a halt. But at the end of the day, fancy footwork aside, Pure is still just a storage vendor. However, with data creation exploding, data being considered the new gold rush, and less than 1% of existing corporate data being used, then ‘just a storage vendor’, especially one “growing like a bat out of hell”, looks like a very good place to be.
The Fiddly Bits
Pure made a number of announcements at last week’s event, including FlashArray//X family, the company’s newest flash array addition powered by NVMe and NVMe-oF. Pure states it ‘makes everything faster – databases, virtualized and containerized environments, test/dev initiatives and web-scale applications – at no additional cost over FlashArray//M, based on effective capacity.’
Expected to start shipping in June, the new line consists of five configurations, from the NVMe-ready //X10 to the all-NVMe //X90, Pure’s densest, fastest FlashArray to-date. It features up to 3PBs of effective capacity in a 6U chassis, latency as low as 250 µs, and offers up to a 2X performance improvement over the FlashArray//M.
Pure also expanded its Evergreen Storage program with the launch of the Pure Evergreen Storage Service (ES2), pay-per-use on-premises storage. Available for terms as short as 12 months with a base commitment as low as 100 effective TBs, ES2 provides on-demand, enterprise-grade storage at up to 50% less than public cloud and can be onboarded in days and offers customers a fully-elastic, scalable on-premises solution complete with tier-one, enterprise grade features.
Less than two months after announcing the ‘industry’s first comprehensive AI-Ready Infrastructure, AIRI, powered by NVIDIA,’ Pure was back with AIRI Mini, a bundled solution that integrates FlashBlade storage, NVIDIA DGX-1 servers, interconnected with 100GbE switches. “AIRI Mini provides organizations without enterprise-level resources an on-ramp to AI that’s simple, performant and ready to deliver value within days,” said Patrick Moorhead, Founder, President and Principal Analyst at Moor Insights & Strategy, in a canned quote.
Also new is the Data-Centric Architecture, its vision for the future of infrastructure. According to the company, back by a survey it had conducted by MIT Technology Review, a DCA ‘keeps data and applications in place while technology is built around it, and fundamentally transforms data center design with the data as its core element.’
“Our customers aggressively seek to use data to improve their customer experience and outdistance their competition,” said Giancarlo in a prepared statement. “The unprecedented explosion of data coupled with new data center technologies enables Pure customers to build a data-centric architecture that enhances and simplifies their ability to use data for intelligence and advantage.”
DISCLAIMER: Pure Storage looked after airfare and hotel, and while I do not own any shares in the company, I do in a number of others mentioned in this article.