Harnessing the Power of Enterprise Grid at eBay

By Nicole Hemsoth

September 11, 2006

In this Q&A with GRIDtoday, Paul Strong, distinguished research scientist at eBay Research Labs, discusses eBay's cutting-edge use of Grid technologies — from where it began to where the online auction leader hopes it to go. In addition, Strong, who has been and continues to be very involved with the Enterprise Grid Alliance, gives his views on the opportunities and challenges facing the newly formed Open Grid Forum.

GRIDtoday: First, could you give our readers a summary of your background in Grid computing and your duties at eBay?

PAUL STRONG: I have been doing distributed systems work for the best part of 20 years, designing and building solutions, ranging from distributed command and control systems for government agencies to large multi-tier e-commerce systems. This has naturally segued into what one would consider more explicitly Grid computing for perhaps the last six or seven years where I worked at Sun Microsystems on its N1 strategy and product set. The focus there was on creating the horizontally integrated platform that is an enterprise grid in the Enterprise Grid Alliance’s parlance. Clearly, such platforms are heterogeneous, hence the need for interoperability standards and my involvement with the EGA as Sun’s rep on the technical steering committee and as chair of the reference model working group.

I currently work in eBay’s Research Labs, where I drive our medium- and long-term strategy for Grid computing and datacenter management. My work includes researching, developing and evaluating new technologies and driving the vision and strategy for our internal tools and systems development, as well as external engagement with the Grid community to drive industry standards.

 
Gt: Given its rapid growth in terms of number of users, I assume eBay has quite an extensive Grid infrastructure in place. Would this assumption be correct?

STRONG: Yes, this is the only way that we could cope with such enormous growth, while still delivering the responsiveness and availability that our community expects. To give you some idea of or growth, from a standing start in 1995, we have grown to more than 200 million registered users, approximately 104 million items available at any given point in time and 1.2 billion page views a day at present. During this time, we have evolved from a traditional multi-tiered architecture to a massively distributed architecture/infrastructure everywhere, including everything from development and test environments to production databases (scaling through disaggregation/partitioning) and Web servers (scaling through replication). We have both what one would consider traditional grid infrastructure elements and enterprise grid infrastructure elements.  
 

Gt: How is eBay utilizing the aforementioned Grid technologies? Is Grid being used for both behind-the-scenes jobs (e.g., storage, database management) as well as customer-facing applications (e.g., payment processing, the eBay auction platform itself)?

STRONG: We view our entire infrastructure as an enterprise grid on which we run both transactional and compute intensive/batch workloads, and which we will manage holistically. The once fixed (more or less) physical silos within which certain services and certain classes of service deployed are becoming more flexible, logical entities that can be created and resized dynamically in response to business needs.

One example of a more traditional Grid environment is ICE, a sub-grid with over 300 servers that we use to build the eBay.com software platform and deploy it for testing. This has reduced our build times for the entire platform from what once took 10 hours to 30 minutes. ICE has done more than 2 million builds in the last 2.5 years. We also view our core applications, the main Web platform itself (essentially a Java-based SOA platform) and our search platform (massively distributed scatter/gather) as transactional Grid services. These two combined account for the best part of 9,000 servers.   

Gt: Speaking of eBay's platform, how has this, as well as the underlying IT architecture as a whole, affected the deployment and management of the Grid infrastructure? Have you had to create a lot of homegrown applications in order to meet the needs of eBay's unique architecture?

STRONG: The sheer scale at which we operate has driven us toward reducing the number of things we manage (through abstraction) and to automate processes. The core services that comprise the platform, and many of our management tools are essentially homegrown — no one has done this kind of thing to this scale before. The great benefit of homegrown applications is that we are able to build the right hooks into them for deployment, monitoring and management. As we develop the applications and services, we are creating an integrated Grid management infrastructure that allows us to rapidly deploy and redeploy them, as well as moving us toward automated service level management and problem diagnosis  Of course, we leverage many vendor components too, and that is where standards will matter as we move forward — allowing us to replace costly-to-develop-and-maintain homegrown technologies with commercial-off-the-shelf components once their functionality has become more common or has been essentially commoditized.
 

Gt: Looking back, how has eBay's use of Grid and related technologies advanced since you first came onboard?

STRONG: The applications or services were already there and we had a disparate set of tools to manage our complete infrastructure (both platform and services). What happened before I joined, and what brought me to eBay, was the explicit recognition that our infrastructure is an enterprise grid. It needs to be managed holistically.  We already leverage the innate scaling and resilience of Grid computing, but we need to make it more manageable, more responsive. So now our challenge is to provide fully integrated, intuitive, service-centric and business-focused management of our infrastructure. Since my joining, we have continued to make great progress in this area. We have leveraged the EGA Reference model. Its classification of the objects we have to manage, their life cycles and the relationships between those objects has heavily influenced our integrated grid management framework, which is now used to manage aspects of our test/QA and production environments.
 

Gt: Looking forward, how do you see (or want to see) eBay's use of Grid technologies evolving in the within the next year or so? What about longer term?

STRONG: We obviously want to continue the Grid-ification trend, for both transactional and compute-intensive applications and make them ever more scalable, agile and manageable. Over the next year, we will continue to build our grid management tool, extending and refining the object model and APIs and using it to orchestrate more and more tools, automating many of the highly repetitive tasks that our operations staff has to undertake. This reduces cost and risk, and improves agility. We already roll code to almost all of our infrastructure every two weeks, which is amazing, but we want integrated and even finer-grained control across all of the elements — compute, network and storage.

In the longer term, we hope to include some of the work that we are developing in eBay Research Labs right now.  We want to query stores of relationships, derive topologies and dependency graphs on the fly, and use inference engines to help diagnose problems. Once we have this level of understanding, and combine it with automated controls, we will be able to deliver greater agility and responsiveness to business need, combined with unparalleled reliability and efficiency. Obviously, this will take time, but we aim to get there one use case at a time.
 

Gt: Moving on to the world of Grid in general, you're quite involved with the Enterprise Grid Alliance. Can you give a background of your position within the organization?

STRONG: Before I joined eBay, I was chair of the technical steering committee and of the reference model working group, where I was lucky enough to work with a great group of people to develop the EGA Reference Model.   
 

Gt: How does work with the EGA affect the work you do at eBay?

STRONG: As I mentioned earlier, the EGA Reference Model has significantly influenced the development our management framework and the way we visualize our grid. We already have UML for parts of it and Java classes.  We look to extend these and to add other technologies to represent and allow us to navigate the relationships between the components that comprise our grid.
 

Gt: Will you still be involved with the EGA now that it has merged with the Global Grid Forum to create the Open Grid Forum? If so, in what capacity will you be participating?

STRONG: Yes I will be involved, at a minimum participating in one or two working groups.  I did have a nine-month hiatus while I settled into eBay, but I expect to give voice to the enterprise-end-user perspective within the new organization. .
 

Gt: Overall, what are your thoughts on the merger? Do you think the new organization will be able to better address the user community's needs in terms of standards and interoperability?

STRONG: I think that the merger was inevitable, there being too many things in common with respect to many of the underlying ideas and technologies that address Grid computing in both the technical computing and enterprise domains. Plus, a bifurcation of the community was both costly and confusing for many.
 

Gt: How do you think the new organization will handle the balance between academic and commercial needs, as well as the relationship between these two sectors?

STRONG: This is clearly the big challenge. There is a vibrant technical computing community that is busy developing the standards and interoperability specifications that meet their needs. The OGF must ensure that they maintain their momentum, that the new body still delivers the value that made the GGF so compelling to this community over the last few years. The enterprise space is subtly different in that the consumers of the technology that incorporate interoperability standards are not, in general, the developers of applications and tools that will realize those standards. Plus, their focus is, in many ways, more on some of the manageability aspects of grids. This makes engagement much harder. But we need that engagement so that commercial organizations can realize the full value of Grid computing, without the fear of vendor lock-in or of tools that aren’t around for, or aren’t supported, over long periods of time.

eBay is clearly on the leading-edge in terms of our applications and services, and we do build a lot of our own applications and tools, so engagement is in some ways easier and more compelling for us — although finding the time is hard when your business is running on “Internet Time.” The OGF needs to find ways to encourage more businesses to engage, either directly or perhaps through advisory councils, in much the same way that vendors engage with customers when deriving requirements for products. The barrier to participation has to be more or less zero, to get the good people away from directly making/saving money for the quarter. Clearly, enterprise use cases need to be reflected in interoperability standards that meet the shared needs of both communities. Some needs are not shared, of course, and some are unique to commercial, enterprise grid users. This will also need to be addressed. But we have a lot of common needs, thus the need for a single forum.

Clearly the next year will be critical. The OGF has a lot to do.
 

Gt: Finally, you're going to be giving a keynote address at GridWorld. What will you be discussing in this presentation?

STRONG: I’ll be discussing eBay’s infrastructure and services as an enterprise grid, how it evolved, where we are and where we are going. Think of the first few parts of this interview as the “appetizer.” Of course, the “entrée” will have more details and pretty slides.
 

Gt: What are your thoughts on GridWorld, in general? In your opinion, what does the ability to host a large conference/trade show incorporating previously independent events (GGF, Globus meetings) say about the state of the Grid community?

STRONG: I’m really looking forward to this year’s GridWorld and the official launch of the OGF. I think the global Grid community — academic, government and commercial users, and vendors alike — needs a single focus from time to time so that we can all see what we’re up to, what’s new, where is there overlap, where there are opportunities and so forth. Distilling makes it easier for everyone, allowing us to maintain critical mass, keeping the costs down and thus making it easier to participate.  

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