Data Centre Energy Savings

Intelligent Products For Data Centers

Artificial Intelligent Software for Data Center Cooling Energy Savings and Condition Monitoring

Clean Energy Products Ltd is happy to announce it has partnered with AdeptDC, an innovative Silicon Valley company that has developed an exceptional Artificial Intelligent control system for data centers. The software enables condition monitoring of the data center servers with a predictive alerts regime and real-time minimization of the data center cooling cost. In effect, the software provides reduced cooling cost combined with reduced server outages.

Data center operations managers strive to maintain critical asset performance such as CPU/memory/disk utilization, network latency, and storage throughput at minimum operational costs which also includes the energy cost of the server cooling systems.

However, the server cooling requirements and the energy usage associated with the electrical chillers vary substantially throughout the day and night as server usage  and hence electricity demand fluctuates. Also, cooling energy will vary as the ambient dry bulb and wet bulb temperature changes depending on the chillers and condensers in use. Free cooling using economizer loops and chiller bypass systems may also be possible during certain times of the year and even at night-time. Cold storage systems such as ice banks can also be used for peak times may also be integrated into the data center cooling systems for improvement to the center’s overall security of cooling supply.

We offer specialized services to data center’s that are aimed to minimize your data center’s cooling cost with typical savings of 40% of your energy usage while using the artificial intelligence software that will ensure that your data center maintains the energy savings improvements in the long term.

First, we can provide you with a detailed audit of the cooling systems and the potential improvements that can be achieved at low cost, medium cost and higher costs. This study normally takes 3-4 days depending on the cooling system size and complexity.

Second, we can provide the AdeptDC software to allow for continuous cooling system optimization and condition monitoring of your server vital signs. The software installation takes a day or less.

Talk to us today for more information on our data center services.

Whether a performance problem is caused by environmental issues

A Few examples of performance degradation’s caused by thermal issues are:
o Heap memory utilization goes down with increased temperature.
o Core switch latency goes high with increased temperature.
o CMOS battery failure causes booting problem.
o CPU frequency goes down with lower circuit voltage
o Disk IOPS goes down because of fan failure.
o Load balancer slows down with high temperature fluctuation.
o CPU frequency goes down due to high temperature.

If a performance problem is indeed caused by a thermal issue, what is the most energy-efficient mitigation sequence?

•In the absence of a actionable insight, data center operators often indiscriminately blast  cooling air through aisles wasting valuable operational resources. Worse yet, the increased cooling in some cases exacerbates the performance problem.

How to reduce cooling costs without performance degradations?

AdeptDC’s AI software addresses these three big questions for data center operations managers.Its learning-based pattern recognition software quickly identifies anomalous component performance and rapidly determines if this problem is indeed caused by thermal issues.

Thereafter, our data-driven recommendation engine quickly identifies the optimum mitigation sequence. Furthermore, it can even autonomously adjust cooling setpoints to lower cooling energy consumption.

The potential improvements of using AdeptDC with your data center

25X

Better Granularity

10X

More Agility

35X

More Agility

Concerted Operations

Our software solution empowers data center operations with predictive asset management and infrastructure optimization.

Granular Healthcare

Keep your finger on the pulse on your data center in real-time.

Predictive Asset Management and Efficient Cooling

Predict failure modes and reduce cooling waste.

Operational Agility

Avoid sensor installation and stay agile

Problems We Solve

  • Continuous Health Assessment

Get answers to: Is my data center running too hot? What are the high-risk critical assets in my data center now? Can I easily identify the comatose servers in my data center?

  • Predictive Asset Management

Get answers to: How should I balance my computing load? How should I respond to emergency situations? What is the risk distribution? What will the temperatures be tomorrow?

  • Infrastructure Optimization

Get answers to: How can I run more applications with the same infrastructure capacity? What is the best way to achieve more operational efficiency? How can I run more IT applications without increasing cooling capacity?

OPERATIONAL EFFICIENCY USING AI

How It Works?

Our machine learning-based software application processes chip-level dynamic thermal data for granular, low-latency, and power-efficient cooling automation. This strategy provides fine-grained visibility into the thermal health of heterogeneous IT assets and ensures a proper alignment between cooling supply and demand. It can precisely predict IT component failures for predictive health maintenance and automate cooling optimization.

Applications:

  • Get comprehensive health insurance for critical infrastructure.
  • Avoid expensive operational downtimes.
  • Improve capacity utilization.

Core Benefits

  • Low Maintenance

Data center maintenance can be an intractable problem with the unpredictable nature of data center operations and complex thermal response patterns. AdeptDC removes this uncertainty and provides

data center managers with much needed peace of mind.

  • Lower Hardware Mortality Rates

Most hardware failures occur because of temperature, voltage, power, or humidity-related issues. AdeptDC’s solution reduces hardware mortality rates with its predictive analytics and autonomous emergency recovery.

  • More Performance per Watt

Because data center cooling demand is difficult to predict, data center managers are forced to overcompensate their cooling capacity. AdeptDC’s automated cooling optimization removes guesswork from the cooling controls and makes it more efficient.

  • Fewer Unplanned Downtimes

As much as 50% of data center downtime happens because of thermal issues. AdeptDC’s data-informed predictive maintenance and capacity optimization reduces the risks of these downtimes.

  • Higher IT Productivity

Although most data center managers are trained IT professionals, a significant portion of their work hours are spent on calibrating cooling systems and fixing thermal issues. AdeptDC’s predictive maintenance application aims to save time for IT professionals so that they can focus more on their core responsibilities.

  • Holistic

ADEPTDC IS PREDICTIVE ANALYTICS FOR ENTERPRISE IT OPERATIONS

Differentiators

IncumbentsAdeptDCWhy AdeptDC
Use air inlet
temperature for
controls
Uses onboard
sensors for controls
Cost-efficient
infrastructure
management
Measure only
temperature
Measures
temperature,
voltage, airflow, and
other thermal
parameters
Comprehensive
healthcare
Needs calibrationSoftware-based
self-tuning
Better scalability,
easy deployment,
and low maintenance
Piecewise
optimization
Holistic
optimization
Better performance
per watt

Industry Validation

IT component temperature-based set point control is the best way to optimize data center cooling. The problem lies in the compilation of temperature data from IT equipment of various vendors.

  • Jonathan Koomey
    Stanford University

I agree with the idea of delivering cooling based on measured server temperatures. This gets you closer to the actual performance metric (server thermal health) than using proxies, such as cold aisle temperature, which have significant margins built in.

  • Jim Gao
    Google

Who is AdeptDC

AdeptDC was founded on the premise that data canter operations can be significantly improved with a technology that quickly identify the correlation between IT asset thermal profiles at component level granularity and dynamic resource consumption—such as IT power consumption, cooling power consumption, deployed capacity etc. The improvement comes from maximizing data center capacity utilization with the risk of service downtimes from overheating. AdeptDC was founded in October 2014 after its founder, Rajat Ghosh, completed his PhD from Georgia Tech.

During his PhD, the idea of AdeptDC struck on him as he was trying to build an AI model that optimizes data center power consumption based on IT component temperatures. Following his PhD, Dr. Ghosh took part in NSF funded I-Corps program to validate his idea by interviewing 112 data center professionals. Later in July 2015, AdeptDC was awarded with prestigious NSF SBIR award to build initial prototype for the solution. That helped AdeptDC develop an initial prototype and eventually acquire a paid customer at Arrow Electronics. In August 2017, AdeptDC relocated to the Bay Area from Atlanta after raising a pre-seed round from Emergent Ventures.