Finally, Elastic Computing You Don’t Have To Share

We all want the best of both worlds: a car that’s sexy and fast, as well as safe and affordable; a meal that’s healthy and also satisfying. When it comes to IT, companies that are operating at scale have been telling us for years that they wish they could get the agility of the multi-tenant public cloud, along with the simplicity, consistent performance, and predictable cost of single-tenant servers.

Well, I’m here to tell you that you don’t have to choose between the two anymore. Rackspace today is announcing the early availability of a new product that will change the way the world thinks about scaling production applications. It’s called OnMetal — and it delivers elastic computing that you don’t have to share.

OnMetal offers single-tenant, bare-metal servers that users can spin up or down as quickly as VMs — in a matter of minutes. And there’s more:

  • These servers are accessed through an API, just like VMs.
  • You pay by the minute, just like with VMs. But OnMetal servers are much simpler and more powerful than VMs.
  • They are built on OpenStack software and Open Compute hardware. The API for OnMetal is the OpenStack Nova API. It’s familiar to users of any OpenStack public cloud. And users don’t have to worry about vendor lock-in.
  • OnMetal servers are customized for specific workloads.
  • There is no hypervisor, and no virtualization tax.
  • There is no sharing of metal with any other user.

OnMetal came about through the daily conversations that we at Rackspace have with the smartest engineers and leaders at the biggest and fastest-growing Internet companies. What’s on the minds of these cloud users is not the unit price per instance/hour of compute. Instead they are concerned about more significant costs —especially the cost of inconsistent behavior and performance, caused by noisy neighbors on cloud networks or disks.

These pain points are inherent in the multi-tenant nature of today’s public clouds — including ours. They are driving massive increases in application complexity and computing costs as users begin to grow and scale. Those costs are rising even amid the much-ballyhooed price cuts for units of compute by major cloud providers.

This rising complexity and cost on the multi-tenant cloud is hitting customers in four main ways:

  1. They spend more on engineering time and talent to architect for failure on the multi-tenant cloud, which is complex and hard.
  2. They also spend more on engineering to deal with inconsistent performance, which is even harder.
  3. They spend more on infrastructure, because over-provisioning is one of the major ways to compensate for inconsistent performance.
  4. They spend more through the virtualization tax, which can diminish disk and network performance by 5 percent to 20 percent.

The cloud users we talk with believe, as we do, that virtualization and sharing a physical machine are fantastic tools when you’re starting out, or if you stay relatively small. But those users say that once they start to get big and their traffic rises, colocation becomes more attractive for its simplicity of scaling, consistent performance, and predictable cost.

Those conversations got our product teams thinking about how to give everyone the technology that the world’s leading Internet companies use to scale. Those giants don’t use VMs or off-the-shelf servers for their core operations. And neither should you, if you’re big — or plan to get big.

Our aim was simplicity at scale, with a high ratio of performance to cost. Here’s how we’ve achieved those goals:

  • All of the OnMetal servers are solid-state. There are no moving parts. That means less heat. Less vibration. This greatly increases the mean time between failures.
  • The configurations are highly optimized for specific workloads. We have servers customized for databases, for low-latency caching, and for handling web requests. Our database server, for example, offers 3.2TB of PCIe flash storage, 128GB of RAM, and more than 200,000 IOPS. Specializing the gear not only boosts performance but helps keep costs under control.
  • We believe the network should have consistently low latency. That’s why we’re arming OnMetal cabinets with 10-gigabit networking and minimal network over-provisioning.

The final ingredient in the formula for successful scaling involves a company’s most-precious asset — it’s engineering talent. As you grow fast and get big, you need to think carefully about which IT tasks to handle in-house and which ones to delegate to a trusted service provider. If you work with Rackspace, you can leverage our scaling engineers and DevOps specialists to run OnMetal machines and other components of your stack.

This way, your company can stay fast and lean. You can focus on building your app, instead of swelling your payroll with a lot of engineers to run IT that doesn’t differentiate your business.

One company that uses Rackspace to stay fast and lean is Pantheon, which helps businesses build and run more than 70,000 Drupal and WordPress sites. The Pantheon team has test-driven Rackspace OnMetal. Here’s what CEO Zack Rosen says about it: “We’ve all been trained to think of the cloud as generic virtual machines on demand. But the future cloud will be built with containers deployed across bare-metal servers provisioned via API. We believe Rackspace’s OnMetal service is the future of infrastructure as a service.”

Now, let me admit to something that many of you already know: Rackspace is not a hardware company. We are a service company that listens closely to business customers like Zack and works to make computing simpler for them. We are proud of OnMetal because it addresses a big pain point that we’ve heard about from customers.

We’ve designed OnMetal in an entirely open-source manner meant to encourage emulation and improvement. As it catches on, we believe that the technology behind OnMetal will address many of the issues — around isolation and control and security and compliance — that until now have kept many companies running IT in do-it-yourself DCs or in colo facilities.

It will also make it easier for innovative companies to grow. Both of these developments will be good for cloud users and providers alike.

Where we at Rackspace will differentiate ourselves is by providing managed services on top of OnMetal. We will help fast-growing companies scale their most-precious resource — their engineering talent.

OnMetal will be widely available in late July. In the meantime, we invite you to learn more about it, and sign up for a test drive, at We welcome your feedback.

Taylor Rhodes served as CEO of Rackspace, responsible for all of the company’s operations in the U.S. and abroad, until May 2017. He excelled in a variety of leadership roles in sales, support, and international operations, and was known for achieving high levels of engagement with customers and Rackers alike.


Please enter your comment!
Please enter your name here