|
NCS Cataolg of IT Services
Real-Time
Statistics of Cluster Activity
The HPC team provides state-of-the-art computational
resources and expertise for computationally intensive research and
training. The HPC team specializes in the design, build, and management
of Linux/UNIX computational clusters and systems. The team handles
all system software and hardware maintenance, troubleshooting, and
repair. The computational clusters reside in secure, climate-controlled
environments specially designed for high speed and high throughput
systems. The clusters use Intel Xeon or AMD Opteron processors running
Centos Enterprise Linux/Red Hat Enterprise Linux with specialized
clustering management and provisioning software. The internal cluster
networks use Gigabit Ethernet and Infiniband, a low latency, high
bandwidth interconnect. The larger clusters employ high-speed
storage systems with parallel, clustered file systems. The systems
directly connect to Georgetown University's main routers via a Gigabit
link. Georgetown University's routers connect both to the public
Internet via a Gigabit link, and to Internet2's IP network via Mid-Atlantic
Crossroads.
Systems managed by the HPC team comply with Georgetown University's
Information Security guidelines and industry best practices. The
HPC team employs firewalls at both the router and host level to
prevent unauthorized access to the computational resources. In addition,
the systems automatically apply security patches on a daily basis.
A dedicated monitoring and logging system watches for unusual or
suspicious activity, and notifies the HPC team when such events
occur.
Shared Clusters
The HPC team operates the new University purchased shared-use cluster.
The cluster is the first system on campus to use low latency, high
bandwidth networking and parallel, clustered storage. Researchers
and students can request time on the cluster.
Two levels of service are available:
- Standard Account—Users purchase CPU hours on the main
system with no restrictions
- Basic Account—Users use a limited set of resources with
restrictions
Managed Clusters
The HPC team builds and manages clusters owned by specific groups
and departments. The HPC team handles all hardware and software
support, including account management, security, application installation,
and hardware replacement.
Collocated Clusters
The HPC team can provide secure, environmentally controlled collocation
space for Linux/UNIX clusters owned by specific groups and departments
who want to run their own systems. In these cases, the HPC team
does not provide any software or hardware support.
HPC Consulting, Training, and Professional Services
The HPC team offers both hardware and software consulting and training
for HPC applications.
Main Applications
on NCS Clusters
RETURN
|
