Update from Gregory Kurtzer on Singularity’s Push into FS and the Enterprise

By John Russell

September 11, 2018

Editor’s note: Container technology is hardly new but it has undergone rapid evolution in the HPC space in recent years to accommodate traditional science workloads and HPC systems requirements. While Docker containers continue to dominate in the enterprise, other variants are becoming important and one alternative with distinctly HPC roots – Singularity – is making an enterprise push targeting advanced scale workload inclusive of HPC. Singularity got its start at Lawrence Berkeley National lab in 2015 as an open source effort and found near-immediate traction within the HPC community.  

Today, Singularity remains open source but is overseen by start-up company, SyLabs, headed by Gregory Kurtzer who led development of Singularity at LBNL (see HPCwire article, Singularity HPC Container Start-Up – Sylabs – Emerges from Stealth). SyLabs offers a supported version of Singularity and, as mentioned, has turned its attention to the enterprise. In conjunction with the HPC on Wall Street conference taking place this week in New York City, HPCwire asked Kurtzer for an update on advanced container technology progress in the enterprise with a focus on Singularity’s progress and the specific needs of financial services sector, which has long been a leader in adopting HPC and advanced scale systems. 

HPCwire: Greg, maybe provide a quick refresher snapshot of container technology and distinguish it from VMs, and touch on the important characteristics of HPC/advanced scale containers.

Gregory Kurtzer: HPC and what we call EPC (Enterprise Performance Compute) focused applications require direct integration with the host operating system and underlying hardware. In contrast, services require isolation from each other and from the host’s resources. Singularity is designed to mitigate security issues when running containers as non-root, and thus blur the line between host and container. Singularity is also highly performant on both startup and teardown and will completely get out of the way, allowing the application to execute without any interference from the container runtime.

HPCwire: Filling out the container ecosystem a bit, explain the role of deployment/management tools such as Kubernetes or other tools such as job schedulers.

Gregory Kurtzer

Kurtzer: Job schedulers have been part of the core ecosystem on HPC for a long time. Now, with the advent of containers on HPC, we are seeing both HPC and EPC users interested in deployment tools, such as Kubernetes/Kubeflow, for their compute workloads. But the container ecosystem demands more than what a job scheduler offers; now users want to build their containers asynchronously, store them in a clear and verifiably reproducible way, and run their workloads on a different set of resources across multiple resources. These tools now play a big role for users trying to move their workloads between onprem and the cloud.

HPCwire: Looking at the traditional HPC environment (academia and government) and enterprise and financial service sector in particular, how do the container use patterns/needs differ and where are they similar?

Kurtzer: In academia and government, the HPC ecosystem consists of a wide variety of use cases, ranging from massively parallel tightly coupled MPI based applications to single threaded, interpreted based workloads (scripts). The EPC and financial services sector tend to generally have more serial based workloads, but given that all of these are focused on performance and compute, many of them have a very similar nature to the needs of academic and government HPC.

This means that the financial sector resources may not have InfiniBand for interprocess communication, or a parallel file system, but they do need low latency networks usually with TCP offloading for HFT requirements. Additionally, they need to manage unprivileged users and resource management with scheduling as a traditional HPC resource would, but they also need to handle orchestration, CI/CD, DevOps, policy compliance, and change management with validation of their workflows. Singularity is uniquely designed for this.

HPCwire: Security is an issue that is often flagged with Docker and one that’s obviously important in the FS world. What are the security challenges container technology presents and how are they best solved?

Singularity is unique in how it handles security, privilege, and user access. Designed to allow untrusted users to run untrusted containers but in a trusted way, Singularity allows non-root users the ability to run containers while locking their privilege within the container. Singularity actually blocks privilege escalation attempts within the container, so from a security point of view, it is safer for users to run applications from within Singularity.

This use case is not limited to just HPC. Imagine being able to run any service through a container without ever being root, and mitigating to ensure that anything running inside that container can never become root.

But security is also rooted within trust. Singularity (as of version 3.0, which is slated to be released in early October), can support cryptographically signed containers. This means you can trust your container runtime environment. Furthermore, with the Sylabs keystore cloud service, you can verify containers and provide accountability back to the developer. Coupling that with the ability to revoke keys means we can limit the “blast radius” of a given exposure.

Between our security model and trusted environments, Singularity and Sylabs adds an entirely new layer of security to your existing environments.

HPCwire: What are some of the dominant HPC-related container use cases in FS you are seeing?

Kurtzer: We are seeing Singularity’s use cases in EPC and the financial sectors gaining momentum on data driven analytics, simulation, HFT (high frequency trading), and starting to see an uptake in AI for market prediction.

HPCwire: When we talked a little over a year ago, you said Singularity had only a smattering of commercial users. How has that changed and why?

Kurtzer:We have a public “Singularity Registry” in which people can voluntarily list their computational resources which support Singularity. Unsurprisingly it consists almost exclusively of academic and government non-classified systems; most commercial users do not list their resources there. So it is somewhat of a surprise to us every time we are contacted by a commercial organization that is already using Singularity. At this point, we have been contacted by large number of commercial end-users as well as hardware and software providers who are interested in working with us to satisfy the requests of their customers.

One of the motivators here has been on our release of SingularityPRO. SingularityPRO is a curated version of the open source codebase and thus it is made of 100 percent open source software. But, we give a 2-week priority of security updates to paying customers as well as make available the Sylabs Keystore for cryptographic validation of containers (a freemium service to open source users). Lastly, we offer commercial support, training materials, and professional services to PRO customers exclusively as we do not provide enterprise support on the open source codebase.

HPCwire: Can you comment on the growth of Singularity (and other HPC flavors of containers) in the enterprise broadly and perhaps its segments such as FS and manufacturing, etc.

Kurtzer: Singularity’s growth in the enterprise is like a step function, starting with the low hanging fruit of commercial HPC and EPC which introduces Singularity to these enterprises organizations. From there, we have seen that other groups within the organization are introduced to Singularity and love its ease of use, security differentiators, and the novel single-file container image format.

The biggest barrier so far has been the lack of compatibility with existing resources like Kubernetes; but we will soon be releasing a solution for exactly this, so stay tuned!

HPCwire: How is container technology well suited for FS, and what are the challenges?

Kurtzer: Container technology in general is changing the paradigm about what it means to package and distribute software. Singularity takes this to the next level. For example, the Singularity Image Format is to containers what RPM and DEB files are to source code. Our image format, modeled after the ELF binary format in Linux, offers flexibility, control, and cryptographic verification, and thus guaranteed immutability. This changes the DevOps paradigm and offers bit for bit reproducibility and trust.

But for financial services, there are other requirements that have historically made containerization a non-starter. For example, the networking layers of other container systems introduce too much latency for HFT and distributed workflows. Fortunately, Singularity, being designed for performance and compute, does not suffer the same outcome as Singularity does not introduce additional latencies in the network, memory, or IO subsystems.

HPCwire: What kinds of forthcoming container technology changes will help FS segment?

Kurtzer: Security is key as is trust via cryptographic signing and encryption. Additionally, integration with existing resources (Kubernetes, Kubeflow, Mesos), performance metrics, tighter integration into existing workflows, and leveraging a community with an already strong knowledge of the container space.

HPCwire: What’s ahead for container technology, in particular for HPC-capable container technology.

Kurtzer: We see PMIx being a big help for MPI and we are working with Mellanox to solve the OFED compatibility layer.

In terms of the container ecosystem, we are soon to release a Container Library and Marketplace for purchasing of premade containers. This will provide software providers the ability to have software vendors shipping their applications as signed Singularity containers.

HPCwire: It seems like heterogeneous architectures (CPU-Accelerator) are becoming a fact of life with processors diversity in particular – Intel, AMD, IBM/Power, ARM, RISCv – further complicating platform selection and use. How does this affect container technology and what challenges (port to processors, etc.) does it present? For example, which processors does Singularity support.

Kurtzer: There is no limit to which platforms can run compute-driven analytics. Because of this, we support everything possible, including GPU and interconnects. Also, the Singularity Image Format (SIF) contains metadata about what architectures it is built and optimized for such that orchestrators can easily glean insights from the container image itself for orchestration.

HPCwire: Do you expect a convergence between varying flavors?

Kurtzer: No. Options are good as each solution offers variety. There will always be Gnome and KDE, Vim and Emacs, Perl and Python, Docker and Singularity.

HPCwire: You’ve talked in the past about how containers potentially represent a paradigm change in the way applications are delivered; how is this playing out and what should we expect?

Kurtzer: 451 Research believes the application market space to grow to $1.6B in 2018 alone. If the market is growing that rapidly, support of the market needs to be there. Sylabs will be there to support it in terms of technology enhancements and services, including working with other open source projects to create solutions to cover a broad spectrum of customer problems.

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!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues 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 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…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion 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…

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