Computation on Demand: The Promise of Dynamic Provisioning

By By Luke Flemmer, Managing Director, Lab49

December 10, 2007

The concept of hardware virtualization — essentially the decoupling of the execution environment from the hardware substrate on which it runs — is not a new one. The term has been used to describe various such approaches since the 1960s. However, over the last couple of years, a particular flavor of virtualization has started to garner increased attention from progressive IT organizations.

This approach, known as server virtualization, refers to running multiple operating system instances on a single machine, without the need for a host operating system. This approach allows for performance of the virtual operating systems that is very close to the capabilities of the underlying hardware (i.e., virtualization overhead is very low). This removes what was a traditional objection to virtualization — namely that it suffered from the poor performance inherent in software emulation.

Virtualization is one aspect of the increasing sophistication of computational models. Along with compute grids (distribution of computational burden) and data fabrics (the associated distribution of data) it represents an increasing level of abstraction of the computational process. In a sense, it is part of the same continuum that has moved development from register level to compiled programming, and from there to high-level languages built on top of their own virtual machines. In all cases, the drive is toward a more generic and powerful abstraction that is not limited by the intrinsic characteristics of the local environment.

In more pragmatic terms, server virtualization has seen most of its recent adoption driven by its value in addressing a few primary areas: server consolidation, high availability, reliability and testing. Server consolidation refers to the recognition that, in many cases, organizations dedicate specific hosts to running key services. In some cases, this is because the service requires a particular operating system version or configuration; in others, it is to avoid potential interference with critical services by other processes sharing operating system resources. Over time, this can lead to a proliferation of dedicated hardware, all of which is lightly loaded. The ability to consolidate this hardware, using virtualization to provide discrete operating system instances, has proven to be low-hanging fruit for many organizations.

Benefits of high availability and reliability are related. The ability to manage applications as an entire operating system instance that can be quickly restored in case of failure, and the fact that these instances exhibit complete logical isolation even when sharing the underlying host, allow for improved manageability and recovery time and decreased failures due to unexpected operating system-level interactions.

Testing is an obvious use case, and one that is not limited to server virtualization. Even virtualization solutions that use emulation to run the guest operating system on top of a host operating system are extremely useful for testing. The ability to pre-configure operating system instances that can have software installed on them for testing, and then readily discard and re-instantiate them, has been leveraged by testing organizations for several years.

What is lacking in many virtualization efforts, however, is a high-quality provisioning model. It is one thing to take a large number of existing servers and consolidate them through virtualization. It is quite another to leverage virtualization to achieve the promise of on-demand computing.

Last year, Amazon.com launched a new service called EC2, which stands for “Elastic Compute Cloud.” EC2 represents Amazon’s effort to bring server virtualization to mainstream developers — and it is an impressive achievement. EC2 provides a complete model for on-demand computing on a broad basis. Users are able to boot pre-configured operating system images created by Amazon, other vendors or themselves. All instances have both public and internal IP addresses, and traffic over the internal network is both free and fast. The billing rate is per hour of instance uptime, based on public network traffic, and represents only a modest premium over what one would pay for a dedicated machine at a hosting vendor.

Amazon currently provides three logical instance types, evocatively named “Small,” “Large” and “Extra Large,” and ranging in power from the equivalent of a 32-bit 1.1GHz 2007 Xeon processor with 1.7GB of memory to the equivalent of  a 64-bit, quad-core 2.2Ghz machine with 15GB of memory, with cost more or less proportional to computational power. What is compelling, though, is how simple it is for users to scale this computational power dynamically. Once the environment is properly configured, a simple command line instruction can boot or shutdown an arbitrary number of hosts (typically limited to 20, but readily increasable to much higher numbers). For applications that are easily parallelizable — and this includes the bulk of Web applications that scale linearly with their ability to respond to requests — this provides an almost effortless model for managing computational capacity.

For developers working on distributed systems, the experience is even more compelling. No longer does the developer need to worry about provisioning physical hardware to create discrete hosts and to test the interaction between them. Instances can be booted in a matter of seconds, and Amazon provides a simple mechanism for providing configuration data to all instances via a predefined REST (REpresentational State Transfer) scheme. 

Due to security concerns, however, most Fortune 500 companies, especially those that have computation centers based on proprietary data (e.g., financial and pharmaceutical firms), probably will not be able to leverage EC2 directly. While it is possible that Amazon could evolve into a provider that is able to offer the requisite level of security assurance to such firms, it seems more likely that commercial usage will be limited to smaller companies, particularly Web-based start-ups. However, the large firms have a great deal to learn from the elegance of Amazon’s implementation. If they are not already, they should be seriously considering dynamic provisioning schemes along these lines.

For many firms, the capacity requirements of different groups are widely variable. In finance, for example, the load on trading, pricing and risk systems is heavily dependent on the economic calendar or events in the market. Trading activity associated with particular economic announcements can result in computational demands that are two orders of magnitude above those of normal periods. Similarly, for much scientific analysis, usage patterns are characterized by long periods of quiescence interspersed with intensive computation.

For large organizations, dynamic provisioning offers the promise of a statistical smoothing of these load profiles. Instead of adopting the standard approach, which is to specify the hardware capacity at a level that can scale to the peak loads but is severely underutilized the rest of the time, organizations could provision computation services where required from a computational infrastructure that is sized to the mean computational requirements of the organization. Much computation is uncorrelated between different groups, and the dynamic reallocation of services will allow for much more efficient and timely allocation of computing resources.

Further, the aforementioned trend toward abstraction implies a move away from knowledge of the underlying computational hardware. In most large organizations, developers and users of computational systems do not have physical access to the hardware on which their systems run. The systems are maintained somewhere in a datacenter and identified only by host-names. However, there remains a coupling between the logical host instance and the hardware that provides the computational power. Increasingly, this coupling will be seen as archaic, just as the coupling between the business goal and the CPU register used to hold an intermediate value now seems archaic to us. Amazon’s EC2 represents an interesting first vision of this future world.

About Luke Flemmer

Luke Flemmer is a managing director and co-founder of Lab49, a consulting firm that specializes in building advanced applications for global financial institutions, and advising firms on their technology strategy.

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