This post is the first in a series detailing the major pieces and parts of the modern data center, how they apply toward building or upgrading an aging IT infrastructure, and future proofing it for years to come. Modernizing your data center is a daunting task. Most of us really don’t think about the three most basic parts of our infrastructure (racks, cooling, and power) very much, but maybe we should. Modern monitoring tools and practices have come a long way over the years and can give us a ton of data we’ve never had the opportunity to review before. We’re going to consider how better management tools and practices have given us a bigger window into what is really going on in our data center and how that allows us to be proactive in our approach to maintaining our infrastructure.
Racks – Most of us never give racks a second thought beyond how much space we have available for a given project or upgrade, but since we bolt millions of dollars’ worth of equipment to them, perhaps we should put in a bit more thought. The standard 42U rack has evolved. Sure, it looks the same, but tons of features have been added to allow for much greater flexibility beyond just some holes to screw our gear into. We can stick with the tried and true 42U two door rack, but is that really the best fit? Some modern racks are completely self-contained with power and cooling built in, allowing us to create individual climate-controlled environments for our infrastructure and monitor each and every piece of gear inside to a much greater degree than before. If, for example, we are standing up a high-performance web application using a standard old rack with normal power distribution units hanging in the back, we hang the servers, maybe install a switch at the top of the rack and move on. With new self-contained racks, we can monitor the internal temperature and power for each device in the rack. That can warn us when the switch starts pulling a few more volts than normal, possibly pointing to a power supply about to burn out or if the average rack temperature starts to climb by a few degrees maybe a fan is not working properly. These types of alerts can be the difference between a planned hardware swap or a total outage of our application.
Cooling – Sure, we’ve all got huge air handlers installed, often with a couple layers of redundancy, and they make working in the data center an ear-shattering experience. But, what if we didn’t need them? Smart vents and proper layout of hot and cold aisles can greatly increase efficiency and reduce the need for a cooling unit the size of a pickup truck. While it’s likely not feasible for you to swap out all your air handlers, there are many other changes you can make, on a budget, that can increase your cooling efficiency drastically.
- Separate hot and cold aisles – Hanging strip curtains, those plastic strips that overlap and block air, can be a cheap and highly effective way to separate hot and cold aisles from each other and can greatly reduce the effort it takes to cool your equipment.
- Ensure all fans are pulling air from the right direction – This one seems trivial, but every little bit adds up. A few switches installed in the rear of the rack pulling air from the hot aisle can greatly impact the airflow in a rack over time, reducing the life of your gear.
- Block empty rack units – This isn’t just for looks; using blanks to cover empty space in your rack, forces the cold air to go where you want it to go, ensuring that all your devices are getting the proper amount of cold air.
- Manage cables – This is probably the most common issue I see from customers: a rat’s nest of cables covering the back of a rack, absolutely causing airflow issues. Don’t believe me? Just reach in and grab a cable. It’s warm, isn’t it? That’s because its trapping hot air. Hot air must go somewhere and it’s likely going right back into your gear and reducing life and efficiency by making your fans work harder to compensate. Clean up your cables.
Power – Proper power management is a must in any data center. Done properly, it makes it very difficult for a power outage to cause any great harm. However, things have moved beyond the simple dual circuit and into a high-tech realm of informational sensors and trending analysis. The modern data center has power distribution centers that can monitor each individual plug to indicate when a device is pulling more power than normal and can also load balance power across multiple battery backups, generators and even power companies to ensure the most stable power possible. New rail based PDUs can be installed in seconds and are much more versatile than anything in the past. Data-center power management now involves identifying, implementing and monitoring processes to improve power efficiency, not just checking to see if power is on or off. Upgrading and retiring older equipment, replacing it with newer, less power-hungry versions, reduces the overall cost of power consumption while increasing efficiency, density and manageability of the entire data center. New, alternative energy power sources are taking the place of the standard generator and UPS backup combo as well, with some cutting-edge solar and wind-powered systems being installed and touting the ability to go off the grid for days at a time without issue.
While some of these technologies aren’t yet commonplace, they are becoming so very rapidly. When you plan your next data-center upgrade, keep in mind that you should not just upgrade your gear or get a prettier rack, but upgrade your way of thinking and processes as well. These three basic data center components can mean the difference between a nearly maintenance-free infrastructure for your mission, or a costly nightmare which requires constant attention.
When you plan your next upgrade, ask yourself, “Is this the best way to do this?” With a little research, and a quick call to the experts at Swish, we will help answer these questions and many more. You can reach us by phone at 703.635.3324 or by email at firstname.lastname@example.org