640
M. Delfino
2000s, starting a new cycle in the development of outsourced computing services.
The first large scale commercialization was done in 2006, when Amazon launched
the Elastic Compute Cloud (EC2) service. The first large-scale production cloud
services in the academic domain were NASA’s OpenNebula 85 and the services
deployed in the RESERVOIR EU-funded project. 86
Cloud computing is still under heavy evolution, and a full description of the
technology is beyond the scope of this work. It deploys interfaces and toolkits
that allow users to configure virtual computers, storage servers, clusters and even
networks. The ultimate goal is to use software to define the characteristics of a
distributed computing service and then automatically map it to the needed hardware.
This process is called provisioning.
Cloud computing can be used to implement private research data centers. CERN
has migrated essentially all of its data center platforms to be managed by the
OpenStack 87 cloud management system. However, the benefits of introducing
cloud-style management in smaller data centers are not evident at present, as
standard clusters are probably sufficient to fulfill the needs. In addition, there is at
present a lack of personnel trained in cloud computing deployment and operation.
Cloud computing technologies started in the academic world, but are now driven
by the cloud computing industry, which has grown in a decade to have sales of
over 200 G$/year. Cloud infrastructure companies deploy data centers which are
much larger than academic computing centers, with the exception of a few academic
supercomputing centers. Computer room floor areas above one hundred thousand
square meters, power feeds above 50 MW 88 and capabilities for hosting tens of
thousand of servers and close to a million cores are typical.
These huge deployments bring with them large economies of scale, making them
attractive for scientific computing. In some cases, U.S. funding agencies have started
to give grants for the purchase of commercial cloud computing services, replacing
funding for purchasing research computing clusters. Gateways have been developed
to include commercial clouds as part of WLCG, especially for CPU intensive tasks
such as simulations. In the European Union, the HNSciCloud 89 pre-commercial
procurement project in 2016–2018 was aimed to stimulate industry developments
to enable hybrid clouds between research and commercial data centers capable of
85 Nebula Cloud Computing Platform (20 November 2012) https://www.nasa.gov/open/nebula.
html
86 Rochwerger, Benny, et al. “The reservoir model and architecture for open federated cloud
computing.” IBM Journal of Research and Development 53.4 (2009): 4–1.
87 https://www.openstack.org/software/
88 For comparison, CERN’s Meyrin data center power was 3.5 MW in 2018. “Data Centre: Key
Info & Numbers”. http://information-technology.web.cern.ch/sites/information-technology.web.
cern.ch/files/CERNDataCentre_KeyInformation_November2018V1.docx.pdf. Retrieved 3 February 2019.
89 Helix Nebula – The Science Cloud with Grant Agreement 687614 is a Pre-Commercial
Procurement Action funded by H2020 Framework Programme. More information at https://
hnscicloud.eu
M. Delfino
2000s, starting a new cycle in the development of outsourced computing services.
The first large scale commercialization was done in 2006, when Amazon launched
the Elastic Compute Cloud (EC2) service. The first large-scale production cloud
services in the academic domain were NASA’s OpenNebula 85 and the services
deployed in the RESERVOIR EU-funded project. 86
Cloud computing is still under heavy evolution, and a full description of the
technology is beyond the scope of this work. It deploys interfaces and toolkits
that allow users to configure virtual computers, storage servers, clusters and even
networks. The ultimate goal is to use software to define the characteristics of a
distributed computing service and then automatically map it to the needed hardware.
This process is called provisioning.
Cloud computing can be used to implement private research data centers. CERN
has migrated essentially all of its data center platforms to be managed by the
OpenStack 87 cloud management system. However, the benefits of introducing
cloud-style management in smaller data centers are not evident at present, as
standard clusters are probably sufficient to fulfill the needs. In addition, there is at
present a lack of personnel trained in cloud computing deployment and operation.
Cloud computing technologies started in the academic world, but are now driven
by the cloud computing industry, which has grown in a decade to have sales of
over 200 G$/year. Cloud infrastructure companies deploy data centers which are
much larger than academic computing centers, with the exception of a few academic
supercomputing centers. Computer room floor areas above one hundred thousand
square meters, power feeds above 50 MW 88 and capabilities for hosting tens of
thousand of servers and close to a million cores are typical.
These huge deployments bring with them large economies of scale, making them
attractive for scientific computing. In some cases, U.S. funding agencies have started
to give grants for the purchase of commercial cloud computing services, replacing
funding for purchasing research computing clusters. Gateways have been developed
to include commercial clouds as part of WLCG, especially for CPU intensive tasks
such as simulations. In the European Union, the HNSciCloud 89 pre-commercial
procurement project in 2016–2018 was aimed to stimulate industry developments
to enable hybrid clouds between research and commercial data centers capable of
85 Nebula Cloud Computing Platform (20 November 2012) https://www.nasa.gov/open/nebula.
html
86 Rochwerger, Benny, et al. “The reservoir model and architecture for open federated cloud
computing.” IBM Journal of Research and Development 53.4 (2009): 4–1.
87 https://www.openstack.org/software/
88 For comparison, CERN’s Meyrin data center power was 3.5 MW in 2018. “Data Centre: Key
Info & Numbers”. http://information-technology.web.cern.ch/sites/information-technology.web.
cern.ch/files/CERNDataCentre_KeyInformation_November2018V1.docx.pdf. Retrieved 3 February 2019.
89 Helix Nebula – The Science Cloud with Grant Agreement 687614 is a Pre-Commercial
Procurement Action funded by H2020 Framework Programme. More information at https://
hnscicloud.eu