May 16, 2010
One of the most effective means of understanding cloud computing in the HPC context is to look to the multiple use cases that have been springing up with particular velocity over the last several months. From large-scale labs and projects on the international scale to mid-sized enterprises with big data demands, use cases present the proof of concept needed to make the transition to cloud -- often private clouds -- more tangible for HPC. There is certainly no denying that there are challenges specific to HPC when it comes to clouds, thus the best use cases are those that present thorough discussion of some of the challenges and show how enterprises have addressed such barriers.
Finding cloud computing use cases for HPC is not difficult necessarily, but it does take some footwork, especially if one is looking to step beyond the multitude of use cases that are provided by cloud and HPC vendors. While these are often still helpful, one must always reserve some room for doubt since these documents are functioning as sales pieces just as much as they might double as proof of concept validation for a particular provider or technology.
If use cases are the key to presenting a balanced proof of concept for HPC in the cloud, then it's critical that we discover meaningful ways of extracting these practical examples from the hype and hot air of marketing that comes from firms that want to show how their technologies are the perfect fit for cloud adoption. This means that to get unbiased cloud adoption studies for HPC we're relegated to combing through national lab and research institution databases just to get what we hope will be a clear picture of how HPC functions in the cloud space without holding a grain of salt in hand throughout a recitation.
The most effective way to get objective reports on HPC and cloud use cases outside of going directly to the institutions and labs themselves is to find an organization that is made up of all parties who have something at stake in the cloud debate. Vendors, end users, cloud architects, enterprise and research community members -- everyone. This can be time-consuming, so hopefully this article provides a better starting point than the wide net cast by an Internet search.
Before we begin with the list, we have one notable addition that's still in the works at the National Institute of Standards and Technology (NIST). As you may already know, NIST developed the widely-used full definition of cloud computing and has monitored the topic closely. The agency announced in March that it will be producing a portfolio of cloud computing use cases but we will need to stay tuned as this develops. When it does there is no doubt that it will be the first on this list since NIST is one of the most trustworthy organizations for putting together unbiased practical use cases for enterprise, the public sector, and HPC researchers -- not to mention smaller SaaS vendors.
So, with that out of the way, let's dive in. Some of the more prominent resources have richer descriptions but don't let that fool you -- everything listed here can be of tremendous value, but that value is determined by what your focus is.
The HPC Advisory Council's Advanced Subgroup for Cloud -- The HPC Advisory Council has put together some excellent studies about moving scientific and large-scale computing into the cloud and we expect that there will be more from this effort to come over the next several months. Highlights from this resource include "From Computational Science to Science Discovery: The Next Computing Landscape". HPC cloud use cases aside, this is also a beneficial resource for members of the community to gain insight from a number of experts and end users via email.
Google Groups: Cloud Computing Use Cases -- This is a group that offers the balanced perspective necessary to examine cloud computing use cases for HPC users and for those with more routine business needs. The documents that this group produces are tremendously valuable and present several different scenarios and the issues that were faced and then solved. This group is clearly in favor of showing how cloud computing solves problems and challenges thus you're not going to find any "barriers to cloud" here, but if you're already convinced in the potential of cloud this is a wonderful resource with active, engaging discussions dedicated to practical cloud examples.
HPC and Cloud Security Use Cases from The Fraunhofer Institute for Secure Information Technology (SIT) -- This use case for cloud computing and HPC is a bit more specific than some others due to its emphasis on security but since this is such a major concern for large-scale enterprise and the public sector in general, this is a thorough exploration about how security, governance, and all of those other nebulous matters are handled in a practical, functioning environment. This report it is not just security-related, however. It covers SOA, contract agreements among multiple cloud service providers, patterns of risk assessment, and other topics. For the large-scale enterprise this is required reading. Unfortunately, for this one, you'll need to send a great deal of info and promise your first-born -- but again, might be worth it.
Cloud Computing and Virtualization from the DSA Research Group -- This resource for case studies and technical discussions about HPC and cloud is from the DSA (Distributed Systems Architecture) Research Group at Complutense University of Madrid. The organization "conducts research in advanced distributed computing and virtualization technologies for large-scale infrastructures and resource provisioning platforms" and has an ever-expanding cloud focus. This group provides several publications of interest to the HPC community and has a rich section on cloud and virtualization that includes in-depth technical discussions that include practical examples of cloud and HPC.
Research Agenda in Cloud Technologies -- While this is not a specific set of use cases, the literature review provided in this piece cites multiple HPC use cases in the enterprise and research/academia spheres in addition to presenting an in-depth overview of HPC and cloud alignment and weak areas. This is one of those "required reading" pieces and does a thorough job of highlighting specific HPC and cloud problems and highlights in the context of an overarching discussion.
Energy-Efficient Scheduling of HPC Applications in Cloud Computing -- So it seems necessary to warn that this is a rather technical, hypothetical use case of reducing energy costs while running HPC operations in the cloud. However, even though it's not using a specific firm or user group, it demonstrates very effectively (although a bit abstractly) the practical energy and cost-cutting principles that make cloud attractive. The numbers can be altered based on existing datacenter configurations to produce a functional model to base tailored insights upon. If you can muddle through some of the theory (if that is not your strong suit) it provides a compelling example of what everyone means when they say that the cloud is (or at least can definitely be) more efficient for certain HPC needs.
Cloud Storage Use Cases -- From the Storage Networking Industry Association (SNIA), an organization dedicated to storage and networking, this document summarizes cloud storage scenarios across cloud storage vendors using a number of different levels of compute resources. While it is focused on cloud storage and only a few of the elements apply directly to HPC, this source provides users with a way to see how storage choices can have an impact on performance and cost.
Top HPC Use Cases in the Life Sciences from Sun Microsystems (pre-Oracle) -- So yes, as you noticed, this is from Sun, so it's necessary to note that objectivity might be a bit compromised. However, this is an excellent resource for the life sciences as it moves more of its big data processing needs to the cloud. Life sciences (along with financial services) is one of the hottest markets for HPC and cloud, thus one can expect to see far more added to a list of use cases that will spring up elsewhere in this industry. This requires a download but again, as with others that require some effort on your part.
Inc. Magazine's Uses Cases for Everyone -- I know, I know -- HPC isn't just for "everyone" but still, this is a great basic use case scenario that explores the "cloudbursting" model for big data can be found in the example of TC3 Health. What works here is that this example can be twisted to fit just about any other major-data enterprise, thus making it broadly useful and applicable. Although this is not necessarily high-level for those already very familiar with the concept and practice of cloud computing, this piece from Inc. Magazine (even though a bit dated already) provides three great use cases for different types of users. For those who happened upon this list without a firm understanding of how cloud might be leveraged for enterprise purposes, this is the best introduction.
Big Cases, Small Grains of Salt
While we aiming for objectivity with this list, there are some vendors who do have some powerful HPC and cloud use cases, including (but certainly not limited to) RightScale, Adaptive Computing (which has great case studies for the University of Cambridge, Rocky Mountain Supercomputing, the Holland Supercomputing Center, the Barcelona Supercomputing Center, TriLabs, Oak Ridge National Labs, and even the Weather Channel along with CHPC in South Africa), SGI and its discussion of the Cyclone offering for HPC cloud, Platform's use cases for ISF, also, another vendor, UnivaUD has a large selection of white papers that include some use cases that are particularly valuable. A host of others that can be found by visiting each of the vendors listed in this article about cloud vendors for HPC.
Linking In to HPC Use Cases and Discussions
One often-overlooked source for valuable conversations about use cases that is held between end users and vendors is LinkedIn's groups. For those who have never used these groups, you are able to follow multiple threads, correspond with those on all ends of the HPC cloud operation in different contexts (enterprise, academia, public sector), and ask questions about actual use of clouds in the HPC space. Among the most valuable HPC and cloud LinkedIn groups are:
HPCcloud Group -- This is probably one of the most useful of those listed. With close to 600 members (at this time), this group provides a platform for end users and vendors to discuss a wide range of issues related to moving HPC into the clouds. From the highly specific and technical to some of the more ethereal matters the community faces, if you're going to join any of the groups listed here, make this your first stop.
High Performance Computing has around 3,000 HPC professionals and is affiliated with the TOP500 and its similarly-named counterpart High-Performance Computing Group with close to 700 members, mostly industry professionals.
HPC Professionals has around 1,000 members and some more recent conversations from particular members are becoming cloud-centric. This is also a good source for networking if you're looking to discover and build your own set of use cases based on individual experiences.
Cloud Computing, VMware, Virtualization and Enterprise 2.0 Group -- Power in numbers is the name of the game with this particular group; almost 30,000 subscribers share specific cases with particular emphasis on the enterprise. Good source for discussions about practical examples of virtualization and clouds of all types.
Open Cloud Manifesto -- An advocacy group of around 1,600 members with the mission of keeping cloud open and accessible with several interesting contributors, including Reuven Cohen of Enomaly.
Below are some others that are necessary additions to the list but that either have a mixed focus or only update new case studies as they emerge. These are all noteworthy as well because of what they offer to the broader HPC and cloud community, either in terms of mission, research and publication focus, or general relevance. If one is looking to compile a list of HPC and cloud use cases, keeping tabs on the following will be of immense value as new materials are published:
If you're sitting there shaking your head because of a notable omission from this list, send an email or leave a comment letting everyone know about other valuable resources that might have been overlooked. Ideally, we'll see an organization emerge in coming months that is dedicated to compiling and creating HPC and cloud use cases specifically, but until then, we'll need to rely on site-hopping to form a profile for cloud viability in the HPC context.
May 23, 2013 |
The study of climate change is one of those scientific problems where it is almost essential to model the entire Earth to attain accurate results and make worthwhile predictions. In an attempt to make climate science more accessible to smaller research facilities, NASA introduced what they call ‘Climate in a Box,’ a system they note acts as a desktop supercomputer.
May 16, 2013 |
When it comes to cloud, long distances mean unacceptably high latencies. Researchers from the University of Bonn in Germany examined those latency issues of doing CFD modeling in the cloud by utilizing a common CFD and its utilization in HPC instance types including both CPU and GPU cores of Amazon EC2.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/02/2012 | AMD | Developers today are just beginning to explore the potential of heterogeneous computing, but the potential for this new paradigm is huge. This brief article reviews how the technology might impact a range of application development areas, including client experiences and cloud-based data management. As platforms like OpenCL continue to evolve, the benefits of heterogeneous computing will become even more accessible. Use this quick article to jump-start your own thinking on heterogeneous computing.