Executive Summaries for Key Findings of Data Center Research Professionals

Four brand new Executive Summaries for the latest data center research key findings revealed from AFCOM Leaders Lab White papers. AFCOM Leaders Lab is a think tank comprised of data center professionals across the United States who explore solutions and develop responses to the most critical issues facing data centers today.

In collaboration with AFCOM Leaders Lab’s Data Center Institute board of directors, more than 100 industry experts gathered together in four cities to produce educational white papers on the following topics:

Content Summary

Beyond Green: What the Next-Generation Sustainable Data Center Will Look Like [Seattle]
DCIM: Take Back Your Nights and Weekends! [Dallas]
Evolving from Reactive to Predictive Maintenance [New York/New Jersey]
Data Center Optimization and TCO [Phoenix]

Beyond Green: What the Next-Generation Sustainable Data Center Will Look Like [Seattle]

Several surveys indicate that data center sustainability has fallen to the bottom of many companies’ priority lists or completely off the list in some cases. In fact, a new report from Uptime Institute revealed that while PUEs are dropping, outages are also increasing—something that’s going to make selling an IT sustainability strategy to your CIO that much harder. We aim to change that.

With the cost of downtime increasing exponentially, reliability will continue to be king and capacity next in line. However, it doesn’t need to be—nor should it be—at sustainability’s expense. AFCOM’s Leaders Lab participants who met in September 2018 spent hours discussing ways to bring sustainability back into vogue and sharing strategies for optimizing it in data centers without breaking the bank. They came up with some invaluable and simple ways for developing, and selling, a plan for optimization. Here are 5 tips to get you started.

5 Keys to Maximizing Data Center Sustainability

• Find Your Sweet Spot: What’s a Sweet Spot, you ask? It’s the point where the decrease in the air conditioning load, cooling load, and the increase in the server fan load hit a minimum. You can find it by using a simple, low-risk technique that can save 15% or more on cooling costs. Your CIO will thank you when the next electric bill hits the accounting department.
• Get to Know REBA: Take advantage of the Renewable Energy Buyers Alliance (REBA), an organization that helps companies without massive data centers or budgets purchase cheap power. REBA engages in collective bargaining so, for example, three companies that each want 5 MW of solar or wind power can join forces to ask their utility for a 15 MW project. REBA can help make those connections for you.
• Pick the Lowest-hanging Fruit First: Before you start the very expensive task of replacing inefficient legacy systems, eliminate the simple, easy and cheap ways of saving energy. For example, power down unused servers, close up gaps between cabinets and between servers, put doors at the ends of your hot aisles, or use a heat exchanger to take advantage of free cooling in the winter.
• Measure, Measure, Measure: Before you can even think about meeting with your CIO about a strategy for optimizing sustainability, measure everything you can in the data center from a capacity perspective. Create graphs and present real data that speaks his or her language. Numbers talk.
• Don’t Just Open the Windows: If your data center is located in an area conducive to free cooling, you need to be very concerned about, or at least aware of, the potential for outdoor air contaminants. If you pull air into a facility and it has a whole lot of particulate matter in it, it’s going to cause problems. Have a plan for controlling or mitigating pollutants.

DCIM: Take Back Your Nights and Weekends! [Dallas]

Managing the data center is more complex and challenging than ever before. In addition to all of the environmental variables that need to be monitored, such as power, cooling and airflow, you’re charged with keeping your facility up and running 24/7/365. The good news is that you now have access to smarter sensors and better integration with DCIM systems. This means you can start making better decisions based on data for your overall data center health.

When the Leaders Lab participants met in Dallas, Texas in October, they discussed those issues most likely to keep them up at night, specific best practices for implementing DCIM, and which tools and features you should look for in a DCIM solution. You’ll read about those when the full whitepaper is released at Data Center World in March 2019. But, here are some tips for planning and maintaining a successful strategy.

5 Keys for Implementing a DCIM Strategy

• Get Buy-in From Senior Management: If they see this as just another piece of software that requires a renewal or a purchase, management is missing the point. Data center operations tools absolutely impact their business world. They’ll need to know about it and truly understand the importance of DCIM.
• Form a Team Comprised of a Variety of Stakeholders: Line-of-business managers, facilities, and IT should get together and meet regularly. A team on paper won’t do your business or data center any good. Again, the key point here is that these stakeholders need to be active members in contributing to improve overall operations.
• Routinely Review New Offerings: In a data center world that’s beset with constant evolution, an architecture that doesn’t evolve won’t help your business at all. Furthermore, you should “challenge” DCIM vendors to prove that their features work and can do what they say. Proof of concepts are great ways to accomplish this. A good trial and POC can help you experiment with several models, dashboards, reports, and notification systems.
• Have Realistic Deployment Expectations: A DCIM architecture supporting a massive data center operation probably won’t be firing on all cylinders from day one. However, this is oftentimes the expectation. One suggestion was to have new systems running in parallel with existing ones to learn the architecture and fine tune the process. This means deployments will very much be different for everyone and expectations need to be tempered. It’s critical to understand the entire scope as well as know that DCIM is a fluid component of the data center. As the data center evolves, expect DCIM to change as well.
• Verify, Test, Analyze and Repeat: Test your systems, make sure everything is performing optimally, and be sure to document what you’ve done. Remember, gremlins can pop up anywhere in a data center and cause serious issues. These can be cooling challenges, power spikes, unregistered hardware (thanks IT!), and even relocated assets. You need to validate that your DCIM platform is running optimally and has all current information.

Evolving from Reactive to Predictive Maintenance [New York/New Jersey]

Years ago, data center maintenance was mostly reactive. When a server overheated, it would be repaired. That’s all well and good until you’re reacting to an incident that poses safety risk or results in catastrophic device failure and, ultimately, revenue loss and/or higher costs. Then, we evolved to inspect and/or replace components on a maintenance schedule. We eventually learned we could adjust the schedules for routine maintenance cycles based on historical data analysis. Today, data centers can anticipate issues before they happen thanks to technologies such as big data, machine learning, deep learning and artificial intelligence.

Welcome to the dawn of the era of predictive maintenance based on real-time data! While the Leaders Lab participants who met in October went into deep detail about types of predictive maintenance, best practices and reasons the industry must adopt the strategy, here are some tips for mitigating data collection challenges. You’ll have access to the entire whitepaper, set to be released at Data Center World in March 2019.

3 Ways to Mitigate Data Collection Challenges

• Start With Collecting Basic Metrics: Most any device in use today can offer some data about its health or operation. Metrics can be gathered from both IT and facilities, giving a broader operational picture. Some basic metrics include power utilization, HVAC performance, temperature, load, humidity, and so forth. it is important to account for all critical systems and determine what data points are available and how that data can be used. For some organizations, that may mean normalizing the data and looking for patterns or outliers to identify problems. For others, it may come down to monitoring loads and temperatures.
• Use Methodologies for Measuring Disparate Systems: Knowing data is available is only a small part of the maintenance equation. Making sure that data is relevant is the next step and that means useful data must be gathered. Data gathering can be as simple as doing a daily walkthrough of the data center looking for changes, or it could be as complicated as instituting barcodes and RFID tags and recording movement of equipment. The point is to make sure that data is collected for more than just the sake of collecting data. Try a multi-pronged approach to gathering data that includes: physical inventories, walkthroughs, maintenance plans, ITSM tools, security and auditing.
• Analyze Data Properly: Many of the analytical tools on the market incorporate some form of AI and adaptable heuristics, meaning that patterns can be identified and exposed without the need for a data scientist. The key here is to use the data that matters. A baseline should be created first and then data added to it in stages. That way “noise” can be limited, and focus can be shifted to critical systems. That said, dashboards and reports can be the best means for sharing data with other stakeholders and can be used as tools to drive budgeting decisions. However, you should leverage all available data before asking for more funding. Effective communication of analyzed data is also a big part of the equation. That’s because different stakeholders may be interested in or influenced by different trends or data points.

Data Center Optimization and TCO [Phoenix]

Despite constant evolutions in the industry, the job of capacity planning has remained consistent: to ensure that enough data center resources are available to support any and all IT workloads you plan to run. However, how we accomplish this objective is very different today. It requires a completely different set of tools, skills, language, and even frame of mind than it did just five years ago.

Led by Leaders Lab Advisor Carrie Goetz, global director of technology for data center equipment provider Paige DataCom Solutions, a group of veteran data center engineers, architects, team leaders, and facilities managers— 19 strong—met in September to discuss all of the above and more. At first, they focused on how to work together to improve the business of capacity planning. The discussion quickly elevated to the entire business of how planning, facilitating, and managing the modern data center can be improved through greater participation, education, and involvement.

Although the results of the meeting will culminate with a release of a white paper at Data Center World in Phoenix, March 19-22, here are six key ways to update capacity planning efforts for the modern era:

6 Ways to Update Capacity Planning

• Produce a Glossary: lf the major stakeholders in capacity planning, including data center managers and facilities managers, should jointly produce a glossary that lists and defines common terms and phrases. Equally important is the act of producing the glossary and motivating all parties to consider the work they do from each other’s perspective. Think of it as “walk a mile in my shoes.”
• Rethink ‘Data Center’: identify the new and perhaps misinterpreted metrics that directly pertain to data center performance, efficiency, and availability. By understanding what strings can be pulled where, the opportunity to focus on quality of service as a fundamental variable in the planning equation becomes possible.
• Draw Up a New Roadmap: Create a forward-looking map that represents the present and near-future capacity needs of the data center. Use analytics and visual aids as ways to clearly present potential constraints that users responsible for provisioning resources for new workload might face. Share metrics across the silos to better communicate between teams.
• Make Capacity Part of the Company Plan: Take advantage of your time in the boardroom to present a state-of-the-data-center update to company executives while they’re “all ears.” Boast about the progression you foresee with respect to key data center metrics. Introduce key team members so the C-suite can associate faces and names with IT; not just downtime. And, rethink the tools in use, looking for more cohesive products that benefit both IT and facilities.
• Attract, Retain and Support the Right Talent: Create a climate that incentivizes IT staff to stay with the company, avoiding an expensive, high turnover rate. This includes allowing people to test new ideas, even fail, without fear of losing their jobs. Encourage team-building as well by holding regular “state of the union” meetings across all silos to review plans and create action items.
• Step Up to the Podium: Be a technology leader. Show people in your organization that you understand the goals of IT, can communicate them and, to some extent, embody them. Stress to everyone the need to replenish and renew existing skills to withstand the negative impact of a talent shortage. That may not be as obvious to others as it is to you.

Source from AFCOM Leaders Lab