Did you miss a session at the Data Summit? Watch On-Demand Here.
Today, Vultr, a cloud computing provider, announced a new line of higher-performance options based upon AMD’s EPYC line of chips. The technology delivers better and more predictable performance because it’s based upon a dedicated virtual CPU that isn’t shared between instances.
Cloud computing companies are increasingly emphasizing the power of their chips and highlighting the differences between the earlier generations and the latest options. In the past, cloud companies hid many of these details, perhaps because customers turned to the cloud for commodity computations because they felt it would be better not to fret about specs.
That is changing. Google, for instance, announced last year its Tau VMs also built on AMD’s EPYC chips. It celebrated 56% better absolute performance and 42% better price/performance rates. Amazon has been rolling out a series of Graviton chips built around ARM cores and celebrating their speedy evolution. The Gravitron3 chip, for instance, is said to be three times faster than the Graviton2 chip on some 16-bit floating-point calculations common in machine learning chores.
Moving from Intel to AMD
In the past, Vultr focused only on Intel chips and, late as the day before the announcement, bragged on its web site of offering “100% SSD and high-performance Intel CPUs.” These new AMD chips mark their first step away.
“AMD has made tremendous strides over the past few years in regard to tech innovation. It has pushed the design, architecture and performance of its CPUs, which has enabled us to implement even more affordable alternatives to solutions like AWS EC2.” said J.J. Kardwell, CEO of Constant, the parent company of Vultr. (Constant is the corporate name, but most customers know them as Vultr, which is the public-facing brand. Vultr is their main business.)
The company will also continue to build out the machines with flash disks to avoid bottlenecks for disk-bound software. “Our product includes the NVME storage unlike some other clouds so the IOPS performance is going to be significantly better.” explained David Gucker, the Chief Operating Officer.
Vultr estimates that their new instances will be 40% faster than their predecessors and expect that they’ll offer anywhere from 10 to 50% better price to performance ratios over some bigger cloud companies. They’re quick to add caveats as these claims are estimates that depend upon the workload.
The products do not feature as many commodities as they may seem because the cloud companies use different price formulae for extras like data infiltration, and these can be significant. Vultr believes its machines will be well-priced for CPU-heavy computation with little exfiltration and even better priced for projects that ship many bits.
“Bandwidth is definitely the ultimate hidden charge”, explained Ryan Pollock, vice president of product marketing and developer relations at Vultr. “You prototype and Dev test where there’s no scale and then when you ship you get a huge bill.”
How small cloud providers compete
Vultr is part of a growing tier of smaller cloud computing providers that compete against the big established players by offering better prices. Companies like Digital Ocean, Backblaze and Wasabi are just some of the smaller competitors who can offer prices that can be, occasionally, dramatically better than the best known companies.
While these companies are technically smaller regarding revenue, they can still offer a wide reach. Vultr, for example, has 23 data centers around the globe and plans to offer the new instances in all of them.
“We’ve opened six locations in six months.” said Kardwell. “So for users around the world who care about this, no independent cloud provider is doing more to bring its footprint close to worldwide users.”
Indeed, the team from Vultr used the word “independent” instead of “smaller” during their press briefing. These independent cloud companies can compete on price by offering a basic product with fewer extra bells and whistles. While many of the cloud announcements from big firms like Amazon Web Services or Google celebrate big advances in artificial intelligence, the smaller firms concentrate on offering straight-forward computers with attractive performance for the price.
“The vast majority of users are not endeavoring to buy ML as a service.” said Kardwell. “They don’t want to pay that premium and frankly, they don’t want to wade through the complexity. What you’d rather pay for is the most efficiently delivered compute, storage and networking. We will never try to have 200 plus products. We will always make sure we’re delivering disruptive price to performance.”
These stripped down offerings can sell themselves. Vultr targets developers who can shop and price out computation contracts without assistance or complicated agreements. That means no sales team or complex usage contracts with sometimes hidden incentives.
“Constant joins Slack, Atlassian, and SolarWinds on the extremely short list of business-to-business (B2B) tech companies that achieved $100M of ARR without any sales people,” said J.J. Kardwell, CEO of Constant of the time when they reached that milestone about 18 months ago. They’ve since relented and added a marketing team.
The company also likes to emphasize that it hasn’t been following the traditional path smoothed by VCs. Revenues continue to grow and last week the company announced it had reached $125 million ARR without taking any venture capital.
“Our $125M ARR milestone proves that our bootstrapped, product-led approach can thrive without following the traditional VC-funded route,” said J.J. Kardwell, CEO of Constant last week. “Constant is one of just a handful of companies that has reached this scale without a dollar of equity capital raised. “
VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn More