CSTCloud 2.0
Prioritized as one of the implementation actions of China's National Informatization
Plan 2016-2020, China Science and Technology Cloud (CSTCloud) aims to develop next-generation
e-infrastructure for research to serve innovation and Sustainable Development Goals (SDGs).
The initiative was announced by the Chinese Academy of Sciences (CAS) during the Fourth World
Internet Conference. Implemented by the academy’s Computer Network Information Center,
substantial progress has been made on service framework and technologies, tailored use cases
driven by key innovation challenges and SDGs, and multilateral cooperation for interconnection
and interoperability.
The 2.0 version of CSTCloud pools together 52 platforms and more than 1000 research software,
delivering digital services for compute- and data-intensive research with the total capacities
of core network switching, computing and cloud storage reaching 100G (121G domestic bandwidth,
113G international bandwidth), 315 PF and 150 PB respectively. To support seamless services on
top, CSTCloud has realized the integration and cloudification of high-speed research networks,
massive data storage, large-scale computational analysis, research software, scientific data
and other digital resources. To facilitate the migration to the cloud, technologies have been
developed to support integrated network and cloud services such as high-speed virtual private
network for research, unified scheduling and sharing of heterogeneous computing resources,
distributed data storage, and unified authentication and authorization across multiple
resources.
CSTCloud Services
To provide one-stop portal services, CSTCloud has developed technologies for unified
authentication and authorization supporting one million registered users from the research
communities to apply and acquire diversified services with a few clicks. A technology
protocol stack with dynamic aggregated resources was established to support scalable and
flexible cloud services. To facilitate the transfer, storage and analysis of massive data
produced by large research facilities and international flagship programs, CSTCloud developed
integrated solutions for data-intensive research supported by technologies for hierarchical
management and distributed storage of metadata describing EB-level research data, agile
wide-area virtual private network, optimized congestion control algorithm and multi-channel
concurrent transfer protocol.
CSTCloud Use Cases
Concerning community-based use cases, CSTCloud has supported research teams on key innovation
topics ranging from small particles to the big universe. CSTCloud provided virtual private
network transfer and remote data recovery services for LHCONE project of CAS Institute of High
Energy Physics and the Space Science Pilot Project. To support pulsar search and identification
at Five-hundred-meter Aperture Spherical radio Telescope, CSTCloud provided cloud computing
environment and the platform for transferring and distributing massive astronomical data.
As the pandemic highlights the role of science in serving society, CSTCloud strengthens its
support to interdisciplinary research addressing health and other challenges related to SDGs.
China National GeneBank DataBase (CNGBdb) was connected and supported with more advanced data
transfer and application capacities. Racing against the clock to fight COVID-19, CSTCloud worked
with CAS Shanghai Institute of Materia Medica on antiviral drug screening platform, and developed
the parallel distribution module of the high-performance computing application middleware to
mitigate bottlenecks such as unbalanced computing task load and low distribution efficiency.
The platform completed a virtual drug screening and evaluation task in 120 hours, selecting 30
candidate components from around 70,000 compounds. Extensive supercomputing resources (256 CPU
cores, 4TB memory, 5TP at peak) have been allocated to support COVID-19 research at CAS Institute
of Microbiology in various areas such as virus genomes, metatranscriptomes, metagenomics splicing,
pandemic spreading dynamics simulation and sequence data mining.
Towards the Global Open Science Cloud
To support open science with e-infrastructures and services, CSTCloud and its partners proposed the
idea of Global Open Science Cloud (GOSC) during the 2019 Conference of Committee on Data (CODATA) of
International Science Council in Beijing. GOSC aims to develop inclusive, interconnected and interoperable
e-infrastructures to support cross-border research activities and long-term preservation of research
products. A recent workshop in November 2020 connected more than 80 experts from international organizations
and five continents to address topics related to GOSC policies, research priorities, technical challenges
and funding opportunities. With seeds sowed throughout the year, CSTCloud will work closely with CODATA,
EGI, GÉANT and other stakeholders to discuss the framework, develop cross-continental testbed, and design
prioritized tasks with potential use cases driven by challenges from space physics, molecular biology,
big Earth for SDGs and other data-intensive research communities.
Featured Use Stories