SUBSCRIBE NOW
IN THIS ISSUE
PIPELINE RESOURCES

Advancing Peak Performance in HPC Environments


IHME IT staff members were able to use Intel® DCM to create the needed power statistics for every rack and server model with no additional hardware or software.

Staff installed Intel® DCM to gain greater insight into power demand, thermal efficiency, server utilization, and capacity planning. Intel® DCM does not require the installation of any software agents on managed nodes—and therefore does not impact performance.

The IHME IT team was immediately impressed by Intel® DCM’s short learning curve, ease of use, and simplicity of deployment. Within hours, they were able to compile and aggregate actionable, real-time data from its collection of servers. Intel DCM eliminates the need for complex, device-specific configuration, setup, or customization.

The institute's IT team was able to use Intel® DCM to create the needed power statistics for every rack and server model with no additional hardware or software. Intel® DCM also enabled IHME IT staff to implement a power consumption policy, and the solution’s health monitoring feature allowed the team to receive alerts based on custom power and thermal events, which will further ensure uptime.

“We learned a lot from other products about realistic maximum power consumption, but it was only relevant for historical data, it couldn’t provide us the real-time alerting,” said Vern Harbers, Technical Project Manager, Infrastructure IHME, at the University of Washington. “If something goes wrong in the data center, right now, other products couldn’t tell us that. Intel DCM was easy to plug in, and easy to get the data and analysis from our machines immediately. The alerts and power limitations were set up within a day.”

Increasing efficiency, utilization and uptime

Diving deeper into IHME IT staff’s deployment of Intel® DCM in its HPC data center environment, we find that the solution enabled the team to quickly detect and analyze underutilized systems by monitoring CPU utilization and power consumption over time. Typically, the lack of sufficient workload performance monitoring leads IT administrators to purchase more hardware.

Efficient space and power capacity management is certainly an essential part of operating data centers. However, this becomes increasingly difficult when data centers grow in density and complexity, and with no easy way to get granular power consumption details. Using Intel® DCM, the IHME IT team was able to leverage a single solution for power management across all devices in the data center, supporting the multiple proprietary power measurement and control protocols required by different OEMs.

IHME IT staff members were able to use Intel® DCM to create the needed power statistics for every rack and server model with no additional hardware or software. Hence, they were able to better plan and manage capacity and utilization in racks, safely increase rack densities, and delay adding new racks. Additionally, Intel® DCM enabled the IHME IT team to maintain group power capping while dynamically adapting to changing server loads.

Furthermore, without the control and insight provided by Intel® DCM, it is difficult to gain an integrated view of a server pool and incomplete data sets offer at best limited visibility. A recent study sponsored by Intel® revealed that as many as 43 percent of data centers rely on manual research. However, Intel® DCM analysis allowed IHME to identify and redeploy long-term, low-utilization servers.

Today, HPC is being used to train autonomous robots, assist scientists to find new approaches to clean electricity, and help researchers to design next-generation aircraft, automobiles and shipping vessels that are safer, faster, and more fuel-efficient. Meanwhile, we are happy that Intel® DCM is being used to great advantage at IHME’s HPC data center environment at the University of Washington’s colocation facility and foresee a tremendous value for this technology solution across other HPC data center environments that are striving to increase efficiency, server utilization, and uptime.



FEATURED SPONSOR:

Latest Updates





Subscribe to our YouTube Channel