Magellan Final Report - Office of Science - U.S. Department of Energy
Magellan Final Report - Office of Science - U.S. Department of Energy
Magellan Final Report - Office of Science - U.S. Department of Energy
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
<strong>Magellan</strong> <strong>Final</strong> <strong>Report</strong><br />
2.1.4 Hardware as a Service<br />
Hardware as a Service (HaaS) is also known as “bare-metal provisioning.” The main distinction between<br />
this model and IaaS is that the user-provided operating system s<strong>of</strong>tware stack is provisioned onto the raw<br />
hardware, allowing the users to provide their own custom hypervisor, or to avoid virtualization completely,<br />
along with the performance impact <strong>of</strong> virtualization <strong>of</strong> high-performance hardware such as InfiniBand. The<br />
other difference between HaaS and the other service models is that the user “leases” the entire resource; it<br />
is not shared with other users within a virtual space. With HaaS, the service provider owns the equipment<br />
and is responsible for housing, running, and maintaining it. HaaS provides many <strong>of</strong> the advantages <strong>of</strong> IaaS<br />
and enables greater levels <strong>of</strong> control on the hardware configuration.<br />
2.2 Deployment Models<br />
According to the NIST definition, clouds can have one <strong>of</strong> the following deployment models, depending on<br />
how the cloud infrastructure is operated: (a) public, (b) private, (c) community, or (d) hybrid.<br />
Public Cloud. Public clouds refer to infrastructure provided to the general public by a large industry selling<br />
cloud services. Amazon’s cloud <strong>of</strong>fering would fall in this category. These services are on a pay-as-you-go<br />
basis and can usually be purchased using a credit card.<br />
Private Cloud. A private cloud infrastructure is operated solely for a particular organization and has specific<br />
features that support a specific group <strong>of</strong> policies. Cloud s<strong>of</strong>tware stacks such as Eucalyptus, OpenStack,<br />
and Nimbus are used to provide virtual machines to the user. In this context, <strong>Magellan</strong> can be considered a<br />
private cloud that provides its services to DOE <strong>Office</strong> <strong>of</strong> <strong>Science</strong> users.<br />
Community Cloud. A community cloud infrastructure is shared by several organizations and serves the<br />
needs <strong>of</strong> a special community that has common goals. FutureGrid [32] can be considered a community cloud.<br />
Hybrid Cloud. Hybrid clouds refer to two or more cloud infrastructures that operate independently but<br />
are bound together by technology compliance to enable application portability.<br />
2.3 Other Related Efforts<br />
The <strong>Magellan</strong> project explored a range <strong>of</strong> topics that included evaluating current private cloud s<strong>of</strong>tware and<br />
understanding gaps and limitations, application s<strong>of</strong>tware setup, etc. To the best <strong>of</strong> our knowledge, there<br />
is no prior work that does such an exhaustive study <strong>of</strong> various aspects <strong>of</strong> cloud computing for scientific<br />
applications.<br />
The FutureGrid project [32] provides a testbed, including a geographically distributed set <strong>of</strong> heterogeneous<br />
computing systems, that includes cloud resources. The aim <strong>of</strong> the project is to provide a capability<br />
that makes it possible for researchers to tackle complex research challenges in computer science, whereas<br />
<strong>Magellan</strong> is more focused on serving the needs <strong>of</strong> the science.<br />
A number <strong>of</strong> different groups have conducted feasibility and benchmarking studies <strong>of</strong> running their<br />
scientific applications in the Amazon cloud [67, 40, 15, 52, 53, 51]. Standard benchmarks have also been<br />
evaluated on Amazon EC2 [62, 23, 66, 74, 82]. These studies complement our own experiments which show<br />
that high-end, tightly coupled applications are impacted by the performance characteristics <strong>of</strong> current cloud<br />
environments.<br />
7