Dr John Kan Chief Information Officer of A*STAR (Agency for Science Technology and Research)

ExaNet

CXO THOUGHTS
CIO2 ExaNet
Dr. John Kan, Chief Information Officer, A*STAR (Agency for Science Technology and Research)

The Agency for Science, Technology, and Research (A*STAR’s) current network infrastructure was designed and set up over a period of 15 years since the mid-1990’s. There is no segregation between corporate and research traffic, as both are treated similarly in terms of security and transmission, unlike the networks in National University of Singapore (NUS) and Nanyang Technological University (NTU). This means that heavy usage by researchers could possibly affect network performance for corporate users and vice versa. In the early 2000’s, A*STAR’s Councils and Research Institutes (RIs’) IT infrastructure developed independently of each other, and only in the past couple of years there has been a concerted effort to align the various networks, platforms, and connectivity between the RIs in A*STAR, in order to provide a common environment for increased collaboration among the various institutes. In addition, enhanced research collaboration between A*STAR RIs and external Institutes of Higher Learning in Singapore and around the world over the past few years has placed increased demands on network infrastructure to provide seamless and high-speed connectivity for research community.

A*STAR’s researchers today are involved in diverse sets of advanced simulations and applications such as high-resolution medical imaging, molecular modeling, 3-D simulations, telepresence research, and genetic sequencing analysis – all of which require ultra-high-speed networks to allow A*STAR researchers to interact with other fellow researchers and scientists in Singapore, and with partners and collaboration agencies around the world.

In order to achieve seamless collaboration at high-speeds both within A*STAR and with other local/overseas partners, a next-generation high-performance network (called ExaNet) was commissioned in Jan 2015 to cater for A*STAR’s One-North intra-campus connectivity for researchers’ collaboration needs. The 1 st phase has been prioritized for researchers with immediate needs for high-bandwidth applications. The ExaNet is designed to be scalable to cater to all researchers at One North campus in subsequent phases when the demand grows.

ExaNet is a state-of-the-art high-performance 100GE network currently deployed for up to 1000 A*STAR researchers, with up to 10GE connectivity to the desk. This new network enables researchers to collaborate and share large datasets between end users, storage and compute clusters for increased research effectiveness and pave the way for greater scientific outcomes. The ExaNet is connected to local universities (NUS and NTU), and external research partners via the SingAREN Lightwave Internet Exchange (SLIX), which is a 100GE fibre backbone spanning between A*STAR, NUS, NTU, and SingAREN at Global Switch. ExaNet associated 1500 wired LAN switch ports and 60 wireless access points were deployed to various Research Institutes (RIs) and entities of A*STAR. The ExaNet system enables scientific dataset transfer using next-generation data transfer nodes, directly using next-generation data transfer nodes, directly from RI locations in One-North connecting to overseas, to the US at 100Gbps, to Japan at 100Gbps, to Korea at 100Gbps by end Apr 2019, to Europe/UK at 100 Gbps by end June 2019, and to Australia at 100Gbps by end July 2019.

The ExaNet is built on Cisco technologies such as fabric path, virtual device contexts, and virtual path channels. These technologies enable multiple links to be simultaneously active (instead of the traditional active-standby links), thus contributing to higher bandwidth. They also enable subnets to be spanned across long distances, as well as support multiple routing gateways on the same subnet. Virtual Device Contexts (VDCs) enable a single switch to be virtualized into multiple instances such that traffic from one instance does not interfere with traffic from other instances. The network is also designed to be ready to harness the benefits of new technologies such as Software Defined Networking (SDN) and virtualization.

The faster network speeds (up to 50 times) will allow researchers to collaborate more effectively on large datasets and other bandwidth-intensive applications where previously, they were unable to do so or took days to transfer large datasets (in terms of Gigabytes or Terabytes), which was unproductive and impeded collaboration.

The Smart Nation vision is built around three focus areas, i.e. Connect, Collect and Comprehend. The ExaNet forms a key technology enabler in the push towards a Smart Nation, whereby the ability to Connect via high-performance networks between the sensors that Collect large datasets and compute resources that Comprehend these big data, will form the common technical architecture to support the Smart Nation ecosystem. With the implementation of this state-of-the-art network, A*STAR is well-poised to be a research hub-of-choice among the international research community.