Today, an often forgotten or overlooked aspect of virtual machines is the replacement of IT infrastructure, such as servers with an additional energy-efficient abstracted version of the hardware. This is what has contributed to sustainability gains over the last ten years.
Data centers operating with a massive number of machine learning (ML) workloads pose a risk to offset those gains, encouraging NSF (National Science Foundation) and VMware to come together to contribute toward the digital infrastructure sustainability initiative. And as hardware virtualization grows like better and improved datacenters reach a point of diminishing returns; however, the partners are looking for new ways to flatten the energy curve, and ultimately achieve “carbon freedom.”
This is NSF’s third infrastructure partnership with VMware. It looks forward to shortlisting at least two projects by early next year associated with the Next Generation Sustainable Digital Infrastructure program. As per program officials, financing for sustainability projects with an amount close to USD 3 million can extend to three years. VMware plans to fund a certain amount of toward the project funding.
VMware’s Chief Research Officer David Tennenhouse said, “With machine learning workloads creating “voracious demand” for datacenter power, the aspirational goal of the sustainability partnership is to “knock the [IT power consumption] curve down to get down to carbon neutrality and carbon freedom.”
The primary questions to be addressed in the sustainability effort involved in creating benchmarks necessary for measuring gains and then designing enhanced IT infrastructure and workloads would make the most of the development.
NSF also mentioned, “Prototypes would enable “significant progress toward maximizing the sustainability of the digital infrastructure with minimal impact on programmer productivity and digital infrastructure performance and scalability.”
Additionally, increasing the software-defined system along with cloud-native apps seems to improve the sustainability quotient as datacenters bear the weight of machine learning and various other AI workloads that are operated by power-hungry processors and networking gear.
VMware mentions that server virtualization independently has alleviated data center power consumption by 120 megawatt-hours. Tennenhouse argued that virtualized storage and workloads are efficiently delivered additional energy savings that can support “plateau” the current datacenter power consumption curve.
In due course, NSF is on a spree for fresh and new ideas to curb IT energy consumption as program officials demand researchers to leverage existing tools and frameworks to innovate sustainable methods and prototypes.
Either way, “We do not want the software community to reinvent the wheel,” said Gurdip Singh of NSF’s Computer and Information and Engineering office.
NSF said the deadline for submitting sustainability proposals is November 4, 2020. The agency would then look to transfer the results of the pre-competitive research to enterprise users.