Greater Cell & Rack Power and Cooling Efficiency for Large Racks of PowerEdge Servers

When packing 42U racks to the gills with Dell R760 servers, effective power and cooling control is a necessity. These servers are powerhouses when it comes to performance, but that also means they generate a ton of heat. If you don’t want to experience downtime, maximize efficiency, and protect your infrastructure, thermal optimization is a must. Today I’m going to take you through some of the intelligent steps you can take to manage that thermal load capacity in those racks, all while still maintaining efficient PUE levels. Let’s get to the basics.

Thermal Design Basics

Thermal Design Basics — That’s where we begin. Imagine a 42U rack as something of a small data center in miniature — all those servers piled up on top of one another results in heat, and heat rises and builds up, potentially causing hot spots.

So, what should you know?

• Heat cooks phones from the top down.
• Stuffed with stuff “Dense” or jam-packed shelving units don’t allow air to float between the items, so you’re just increasing the humidity everywhere.
• Each Dell R760 server will require specific cooling depending on the workload and configuration.
• You need good flow from the front (intake) to the back (exhaust) to avoid overheating.

Here’s a good rule: don’t block the airflow paths. Ensure perforated floors, vents and fans are not blocked. Also, do not mix hot and cold air within the rack: airflow direction should be handled with caution.

You may be thinking, “Isn’t cooling just a matter of pumping out more cold air?” Nope. And too much cold air is an energy waster. We are going for best cooling and not the coldest possible. Excessive cooling costs money and, potentially, could be damaging certain components over the long run.

Hot/Cold Aisle Best Practice

You may have heard the term hot and cold aisles, but you can get it very wrong or very right for cooling. Here’s how you can build it for your Dell racks:

• Rack alignment in rows: Inward-facing sides are cool aisles whereas outward-facing sides are hot aisles.
• Cover blank and unused rack spaces with filler panels to minimize bypass airflow and recirculation.
• Consider containment systems, if available. Hot aisles containment or cold aisles containment options for more efficient air distribution, which limits your conditioned air to the area where it’s needed.
• Position CRACS (or in-row coolers) so that they are blowing directly into the cold aisle.

Monitor temperature at various heights in the rack — hotspots are commonly near its top.

In doing so, you create a uniformed environment where cold air roundhouse kicks the servers, and hot air gets captured and removed without mixing. You’ll save cooling energy and protect your costly equipment.

Smart Flow Optimizer in iDRAC

Dell’s integrated Dell Remote Access Controller (iDRAC) tools are excellent, and that includes the Smart Flow Optimizer, an awesome thermal management tool.

Here’s what it can offer you:

• Automatically adjusts fan speeds for each Dell R760 server according to workload and thermal requirement.
• Stop fans running at full speed all the time, save power and reduce noise.
• It has real-time sensors in the server to respond in real time.
• Allows policy setting for temperatures thresholds so it reflects on your cooling.

You should enable Smart Flow Optimizer as it provides you fine-grained control of cooling inside of the server. It is smart in balancing both performance and power consumption.

To do this, simply open the iDRAC interface, navigate to the Thermal Management section, and toggle the feature to the ON position. Watch what happens when it dynamically adjusts fans on the dashboard. In many cases, you can actually have many servers that don’t require full cooling all of the time.

Power Capping Profiles

Power capping is another tool in your optimization toolkit. Just like R760, PowerEdge servers support power restrictions where they will never go over a particular power value. It comes across as impeded performance? Not really.

Here’s why it matters:

• Power capping prevents overloading of your power distribution units (PDUs).
• Eliminates unwanted power spikes that cause breaker time-outs and other unwanted down time.
• Assists keeping total power used by the data center under control for billing, planning, etc.
• Partners with heat controls well enough — experience less power draw = less heat created.

Playing power capping and thermal monitors tight is crucial, as the less power a server consumes, the easier it is to cool down — and the easier on your HVAC system you become.

You can define custom profiles (in iDRAC or OM tools) for preferred max power during various time-of-day workload periods or SLAs. For example:

• There is a possibility that production hours allow a higher power cap to 90% to achieve max performance.
• Can dial down to 70% off-hours for cost savings.
• Aggressive lowering is triggered on emergency profiles if temperature sensors pass critical levels.

Imagine power capping as an intelligent throttle that helps keep your servers running efficiently—and without surprises.

ROI of Efficiency

So now why go to the trouble to make the facilities so power and cooling efficient? The reason is that the R.O.I. is large: in cold, hard cash and in system reliability.

Here’s the bottom line:

• Reduced operating costs: The less power waste, the lower the power bills. A data center’s use of power is typically split 50/50 between compute energy usage and cooling energy usage. Every kW saved matters.
• Less server failure: Server failure is largely due to overheating. Efficient cooling system means less repairs and replacements.
• Improved performance consistency: Servers can operate cooler for more consistent performance under load without the risk of thermal throttle.
• Long service life: good cooling extends the life of hardware. A longer working R760 is a more efficient investment.
• Cyber security advantages: Stable hardware decreases the likelihood of unexpected crashing, keeping your security services active and your threat awareness high.

These are the intangible gains many companies fail to consider. But in today’s security-minded world, uptime is a safeguard not only of data but also of reputation.

Containment systems or iDRAC setup is probably something that’s going to require you to pay something up front, but the savings come back fast. And then, if you can manage load more accurately, that means you don’t have to over-provision your cooling infrastructure either.

So if you have those thick Dell R760 42U racks just keep in your mind:

• Learn fundamental concepts of thermal design including airflow and how heat flows.
• Use hot / cold aisle best practices to isolate air flows.
• Leverage iDRAC with Lifecycle Controller’s Smart Flow to dynamically manage fan speeds.
• Manage power draw and heat generation with power capping profiles.
• Focus on the ROI – Determine where you can cut costs, how long your hardware can last and increase uptime.

If you do things right, your racks’ power and cooling won’t just keep the servers happy – it will safeguard the future of your infrastructure. Begin tuning today for a cooler, safer more efficient data center.

Date: 4/22/2021 Key Words: Power and Cooling for Compact PowerEdge Servers