Any resource is only as good as its infrastructure and thats something we understand all too well. In 2014 we constructed a new Data Center (DC) in the Center for Life Sciences (CeLS) to ensure we had the underlying support structure to cater for all the compute needs we had and would certainly have in the future.
Designed from scratch by our own team in tandem with the vendors, our DC comes fully equip with temperature control, water leak detection, FM200 fire suppression, proximity card and biometric access, SMS and email alert systems, a redundant auto cut in/out UPS with an hour long rundown time and remote access/monitoring of almost everything.
The DC can take 15 racks comfortably with each rack drawing a maximum of 7kW of power and has 64 1Gbit network ports on a 40Gbit backbone and the DC keeps a sustained 16 degree temperature 24 hours a day, 7 days a week, courtesy of 6 industrial grade FCUs working on rotation. Our servers aren’t lagging behind either. As part of our collaboration with The Singapore Centre on Environmental Life Sciences Engineering (SCELSE), we run a small, albeit powerful cluster that handles all the compute needs of SCELSE’s NUS branch.
High-end servers running Citirix XenServer, a hypervisor platform that enables the creation and management of virtualized server infrastructure, allow the deployment over 300 virtual machines of various operating systems for a vast variety of needs, from testing servers to actual production systems.
To cater for research data backup and archival, a 640TB (raw) storage solution with redundancy is deployed. This system connects both over the internet (via VPN) and the internal private network within CeLS via a 10Gbit internal link to allow fast back up and retrieval by the users as and when necessary. Data from completed projects are then compressed, encrypted (requiring two separate keys to decrypt – preventing any attempts at tampering with the data in cases where research paper data needs to be verified) and then archived for a 7 year period.
We also have a central laboratory storage solution with redundancy which allows data from analysis machines such as sequencers and CT scanners to be written over the network directly to the storage instead of to the analysis machine’s local workstation requiring external media to cart the data off for post processing. This allows for data to be accessed from anywhere and only by authorized personnel.