HPC Startup Backs into Cloud

By Nicole Hemsoth

August 29, 2011

Some could argue that it really does take a rocket scientist to address the limitations of high performance computing systems.

According to one rocket scientist we talked to recently, the problem with HPC clusters for aerospace engineers has little to do with the applications or the real science behind the rockets. Rather, he says the thorny issues are rooted in the limitations, complexity, and general care and feeding distractions that require scientists and non IT-experts to work double-duty as HPC system managers.  

Mike Colonno, a former aerospace engineer from SpaceX and current CTO at Black Sky Computing says that scientists, rendering artists, bioinformatics professionals and those in a wide range of other HPC verticals are spending too much time struggling with high performance computing complexities at the expense of their projects.

Colonno pointed to his experiences at SpaceX, claiming that in his daily routine of refining aerodynamic rocket designs, he was forced to spend 80 percent of his time handling IT-related issues versus focusing specifically on his application. His conversations with HPC users in other fields confirmed the suspicion that this wasn’t an aerodynamics industry problem—it was a problem for those who had to contend with cluster computing in any industry.

In a search for a solution to the HPC usability issue, Black Sky was born, at first with the mission to deliver ready-to-roll high performance cloud environments to suit the needs for the many he claimed were seeking hybrid HPC solutions. In short, Colonno says that now—and definitely over the next few years—the hybrid high performance cloud is? in high demand. He says that customers want to maintain their in-house resources that can be easily, seamlessly burst out into a cloud environment to meet time-sensitive demands without incurring vast IT headaches.

The founders of Black Sky, which include Scott Alexander, a former senior software engineer from PayPal, Colonno, and a third cofounder who had also been with SpaceX, found that the existing cloud computing solutions available from the likes of Amazon and others were not suitable for HPC—and that true cloud HPC couldn’t work without servers, storage and networks all pointing to those specific needs. The infrastructure supporting these environments could not, at least according to Colonno and Alexander, support the heavy I/O demands or provide the performance and price match needed for many HPC applications—so that means it’s time for any serious cloud company to build their own.

And this is where things get rather interesting—and at two different ends of the HPC spectrum. First, the company set about building a cloud offering that was comprised of purpose-built high performance computing gear from top to bottom; hyper-efficient servers, 40 gigabit Ethernet, and a robust storage array. While there’s nothing necessarily remarkable in that alone, it is worth noting that the team decided that if they didn’t build it themselves—from servers to storage to software—they couldn’t produce a truly HPC-ready offering.

For Colonno and his team, this led them into a backwards approach to getting into the cloud business. They started off hoping to find a niche by delivering highly focused, refined HPC cloud solutions that carved out anything unnecessary from the server and storage flank—and suddenly found themselves in the hardware business. The strange thing is that for now at least, that hardware push to support the main cloud objectives has been the source of their profitability while the cloud, called SkyNet (not that Skynet) continues to be gussied up during beta in time for real customers sometime late this year or early next.

At this build-versus-buy juncture, one might think going to a vendor like Dell with its DCS service might be the best alternative since a design tailored for HPC users is something they’ve specialized in designing in the past. Colonno says their experiences with the DCS team were excellent—that they were almost close to considering having Dell build the hardware for their cloud. However, the sticky issue was that Dell will provide outstanding support and assistance during the design process, but that ultimately, solutions developed with Dell’s assistance would be added to the Dell server portfolio.

Seeing the uniqueness of their approach to pure HPC-driven hardware, Colonno and Alexander said they made the decision to shed the third party and retain full control over the project. In other words, they simply designed and built their own hardware to support the efficiency, performance, and manageability layers required just for HPC users. Not only did a cloud company thus become a hardware company—that hardware company went on to deliver HPC-tuned systems that could double as physical datacenter resources or come ready-made to burst customers into the cloud on the server, storage and software stack sides.

Colonno said that the incentive to deliver highly customized HPC solutions stretched across multiple areas for users. First of all, he says that companies that cater to the middle market (consumer IT), yet have an HPC division, often have the expertise but the solutions they offer are not tailored, leaving in features that don’t matter for high performance computing. For instance, on the storage side, ripping out things that customers don’t need but that are standard offerings like hourly snapshots of read-only junk refine the offering. By putting the emphasis on I/O in the design, achieving full integration with one vendor for the server, storage and stack, and stressing density, the Black Sky founders say they found something no one else was offering.

Alexander said that one of the weaknesses of off-the-shelf systems (outside of the fact that too many users don’t understand the plethora of problems that can come after its been powered on) is a lack of recognition of basic HPC needs. And here their portfolio of hardware offerings was born, including the Apollo storage line, which sheds the superfluous and “hub and spoke” problem where users need dozens or thousands of computers reading and writing to a central source at full throttle only to be shot down at the bandwidth level. According to Alexander, volume and throughput are central concerns that are often not addressed with non-tailored solutions.

Colonno pointed to the same paring down process at the server level, pointing to Black Sky’s Hyperion line. He claims it took several incarnations and painful lessons learned to arrive at the fact that by focusing on the bare-bones essentials that balance efficiency with dramatic power, density could be increased and at performance and price point that customers could very easily live with.

The team addressed the middleware “glue” that makes management seamless, whether bursting or sticking to in-house resources. Using a custom blend of open source and in-house developed software originally developed to manage SkyNet, the team claims that they might not offer all the bells and whistles of a Platform LSF or Moab commercial solution, but they hand over everything needed to make system management a breeze—and to allow bursting into the cloud so easy that end users won’t even be it has happened. Colonno claims that one of their target markets, the rendering and visual effects industry, responds well to this, given their need for cloud “burstability,” , high performance and ease of use. The end users here are artists who are application gurus, but lack the IT sophistication needed to tame the HPC beasts of rendering.

Colonno’s humble statement that they’ve stumbled upon a “sweet spot” in computing is food for thought. By starting out with the plan to develop an HPC cloud of their own, their need for hyper-efficiency, density, cooling and storage was driven by their own economic interests. They wanted to run the cloud as cheaply as possible without sacrificing on performance. Therefore, the incentive to build the most cost-effective solution was paramount; the  urgency of the design was important to their bottom line.

The cloud itself could be attractive to those desiring a hybrid solution says Colonno, but the hardware itself also caters to the same requirements. 40 gigabit Ethernet, the ability to work with QDR Infiniband or 10 gigabit Ethernet, and efficiency sound enough that Black Sky is willing to bet its own dollars on it.

The fact that this “sweet spot” turned into a hardware portfolio that drew more interest than the cloud they were building it for was a happy accident—but one the team can certainly accept. While SkyNet is enjoying a productive beta, they have sold over a dozen systems and have a few case studies in rendering and aerospace under their belt already. When SkyNet launches later in the year, the team can put their design to the ultimate test—driving a profitable, efficient, performance-geared cloud that brings new business flocking.  Or so they hope.

Company chief Scott Alexander says that there are still challenges that are “man-made” however, even after some of the hardware challenges are met. Software licensing barriers with many companies being afraid to climb on board with anything that breaks the profitable per-node pricing model as well as general bad press around security and performance are preventing clouds from becoming more prominent in HPC verticals (ummm you’re channeling DFW again) However, he claims that once the software companies get on board en masse and clouds continue to develop a solid reputation, the sky is the limit and the “sweet spot” they stumbled upon could bear delicious fruits indeed.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

Nvidia Appoints Andy Grant as EMEA Director of Supercomputing, Higher Education, and AI

March 22, 2024

Nvidia recently appointed Andy Grant as Director, Supercomputing, Higher Education, and AI for Europe, the Middle East, and Africa (EMEA). With over 25 years of high-performance computing (HPC) experience, Grant brings a Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Houston We Have a Solution: Addressing the HPC and Tech Talent Gap

March 15, 2024

Generations of Houstonian teachers, counselors, and parents have either worked in the aerospace industry or know people who do - the prospect of entering the fi Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire