Computing Infrastructure Information

Department of Computer Science and Columbia University Computing and Network Facilities

The departmental computing infrastructure consists of a data center for computational, storage and VoIP services, one large student laboratory, featuring 30 Linux towers each with 2 Nehalem processors, 8 cores and 24GB memory; a remote Linux cluster with 8 servers with 2 cores and 16GB of memory. Additionally, a Linux compute cluster with 8 servers, each with 2 Xeon E5-2620@2.40GHz processors (12 core) with 128 GB of RAM, and a number of computing facilities for individual research labs. The data center houses more than 200TB of Network Appliance file servers, a compute cluster consisting of a Linux cloud with 32 servers each with 2 Nehalem processors, 8 cores and 24GB memory. This cloud can support approximately 5000 of VMware instances. Other computing facilities include, a VoIP telephone system hosted by MITEL, and departmental load balanced web cluster with 6 enterprise class servers, two online backup servers, each with 40TB capacity, along with supporting servers that provide print services, LDAP, DHCP, DNS, tape backup and software distributions to the department. The data center is internally connected via Gigabit Ethernet and accesses the Internet and Internet2 via the campus fiber network. The University provides 802.11ac coverage for most of the campus.

Columbia University Information Technology maintains Columbia’s commodity Internet connections that include 9 Gbps via Cogent, 2.25 Gbps via Level3. Columbia’s research and education network connections include a NYSERNet NYC and statewide regional optical network (“dark fiber”), a 10 Gbps connection to NYSERNet’s IP network and from there to Internet2 and National Lambda Rail (NLR) packet net.

The research infrastructure has approximately one thousand servers including many Dell, HP, and Silicon Mechanics with Tesla GPU; it also includes several workstations from Apple, Dell, HP and other PC manufacturers. The Columbia Robotics Laboratory (Professor Peter K. Allen, Director) is housed in the Schapiro CEPSR building, where it has approx. 1000 sq. ft. of space. Facilities include: Baxter Research Robot, PR2 mobile robot manipulator, Fetch mobile manipulator, Staubli RX-60L Robot arm, Kinova MICO arm, custom built overhead XYZ gantry robot, Toshiba FMA manipulator, Barrett Technology robotic hand, 2 RWI Pioneer mobile robots, 1 Evolution ER-1 robot, 1 RWI ATRV-2 mobile robot with RTK GPS, Leica HDS -500 and HDS-3000 100 meter range scanners, real-time Imaging boards, multiple high resolution color video and still cameras, 12 graphics workstations, numerous PCs, air compressor, and a variety of special purpose pieces of equipment such as interfaces, machining tools and lights.

The labs for research in image processing, vision, graphics, and robotics contain specialized equipment such as, Baxter Research Robot, PR2 mobile robot manipulator, Staubli RX-60L Robot arm, Kinova MICO arm, custom built overhead XYZ gantry robot, Toshiba FMA manipulator, Barrett Technology robotic hand, 2 RWI Pioneer mobile robots, 1 Evolution ER-1 robot, 1 RWI ATRV-2 mobile robot with RTK GPS, Leica HDS -500 and HDS-3000 100 meter range scanners, and real-time Imaging boards.

Technical Support

The research facilities are staffed by professional systems engineers who are responsible for operating system and network support, miscellaneous hardware and software maintenance, and troubleshooting. These staff members allow individual researchers to avoid spending time on hardware and software problems.

User Fees

The Department of Computer Science charges each research personnel a prorated fee of $2,400 per year for use of the computing facilities in the department. This includes use of all servers, cluster computers, storage, backups, network connections, technical consulting and troubleshooting. All faculty, GRA’s and post-doctoral researchers are charged this fee. The computer user fee funds the operation and equipment of the departmental computing infrastructure, and is charged in accordance with Columbia University Controller Agreement #100988 from Feb. 24, 2006.

The Computer Science Dept, established at Columbia in July 1979, works both in experimental (comprising artificial intelligence, machine architecture, computer networking and graphics, databases, distributed systems and software systems) and in theoretical computer science (algorithms and computational complexity). This research work supported to the tune of approximately $18 million per annum, requires a computing infrastructure within the department, managing and providing shared computing and telecommunications services, e.g. over 1000+ machines plus guaranteed full and incremental recoveries, maintaining public and commercial software licenses for programming and engineering tools; and managing the complex and varied networking for each research group to allow for independence from general traffic. The responsibility for providing these facilities falls entirely on the department. Collecting user fees allows the department to offer high-quality facilities and continuous round-the-clock services at a reasonable cost, since the cost would be significantly higher if each research unit procured services individually via consultants. The computing facilities unit (CRF) is comprised of 5 full time system administrators.