By Greg Ng | Trace3 Senior Principal Solutions Architect
When it comes to selecting a case for our smartphones, we often consider the color, size, and level of protection. Sadly, many of us may spend more time choosing a case for our phones than selecting a data center rack for our IT systems. Data center racks house and protect expensive equipment and valuable data integral to their business. Gartner estimates that the average cost of IT downtime costs an organization is an astounding $5,600 per minute! Since enterprise companies can spend up to millions of dollars for their servers, storage, and switches, protecting these expensive assets should not be an afterthought.
Data center racks offer an unseen protection as they help manage the airflow for the systems they contain. Ideal data center spaces have separate cold and hot aisles to intake cool air into the systems and exhaust hot air out of the systems. Racks sit between the cold and hot aisles, providing a channel for the airflow, usually from front (cold) to back (hot), and a barrier to prevent the hot air from mixing with the cold air. Rack accessories such as side panels, blanking panels, and air baffles help to optimize the airflow and barrier between the hot and cold aisles. Choosing the right rack solution can result in maximizing equipment uptime and performance, increasing the life cycle of the hardware, and lowering the energy costs for the rack and the room.
For many years, data center racks were a one-size-fits-all – 42U height (a U or RU is short for Rack Unit and is 1.75 inches) by 24 inches wide and 42 inches deep. System vendors would manufacture their devices to fit within these confines. However, with the growth of Big Data, AI, and Hyper Performance Computing, a one-size-fits-all approach is no longer sufficient. To meet the demand, racks now come in heights as high as 70U. A 24-inch-wide rack is too narrow for these new computing requirements. When installing a rack with over 40 nodes, a width of 28-30 inches will allow more room for adequate cable management for the 160-200 cables necessary for power and networking. With the proliferation of highly dense storage, CPU, and GPU servers, manufacturers are designing these servers deeper than ever. In a 42-inch-deep rack, these servers protrude into the vertically mounted PDUs in the rear. A deeper rack of 48 inches alleviates many challenges for accessing and servicing power supplies, PDUs, and cables in a shallower rack. Great care must be used when considering the rack size for today’s modern workloads and to accommodate tomorrow’s requirements.
There is a growing shift with enterprises choosing to deploy white-colored racks instead of the traditional black. Studies have shown that rows or pods of white racks can reduce the amount of light required in a data center by 45 to 50 percent compared to black racks. This can save energy costs for enterprises that own their own data centers. Beyond just reflecting and illuminating a room, white-colored racks can also reflect and illuminate the inside rear of the rack 80 percent more than black-colored racks! This makes it easier for technicians to correctly identify systems, cables, and components, speeding up the time to repair and reducing errors.
Data center rack architecture plays a critical role in ensuring the efficient, reliable, and secure operation of IT infrastructure. By carefully considering factors like equipment layout, power and cooling, cable management, and security, organizations can design data centers that optimize performance, minimize costs, and meet their evolving needs.
Trace3 has worked on thousands racks at our state-of-the-art Integration Labs and onsite at our customers' data centers. Rack design, architecture and integration is a crucial process to ensure that your data centers are built in the most efficient, cost effective and energy saving manner. With additional data center integration services, we assist organizations ensure that their rack builds are consistent and scalable, to help their business grow with confidence in their data centers.
Greg is an experienced leader with 29+ years in the Enterprise Data Center environment. He has solutions-based experience and certifications in Enterprise hardware and software. He is a Certified Data Center Design Professional and has completed with distinction the first two years of a three-year Master’s Degree in Data Centre Leadership and Management. He is a proven professional committed to Technical Sales, Project Delivery, and Customer Service.