The modern data center looks very different from those built-out ten years ago. The IT industry has been changing at an exponential pace. According to one of Intel’s founders, Gordon More, the computing power is set to double every few years. Even though this fast pace has slowed down a bit in the past ten years, processing power still duplicates every 2.5 to 3 years.
In 2021, data centers’ importance will expand as the cloud’s growth and the emergence of new technologies like AI and 5G take hold.
To remain innovative and competitive, organizations need to rethink and advance how they use data center technology. A good design is essential to creating and maintaining a highly efficient data center, no matter if it’s small or hyper-scale, used by a cloud service provider or an organization planning to keep their data on-prem.
A good data center design strategy can have a considerable impact on the business, leading to significant cost savings, increased scalability, and decreased environmental impact.
DATA CENTER DESIGN GUIDELINES
A data center’s design comprises a network of computing and storage resources that allow the delivery of shared applications and data. Core components of a data center’s infrastructure include routers, switches, firewalls, storage systems, servers, and application delivery controllers.
The following points help forecast the company’s needs:
- Floorspace – How many square feet of floor space does the data center need today? Do we anticipate the data center room layout to change over time? Do we expect to extend later? It costs less to build what we need now than renovate in a few years.
- Power Requirements – A data center’s electrical requirements can be extensive. We should ensure the data center layout can accommodate future needs.
- Cooling Requirements – The data center cooling design should allow us to increase the number of hardware units without worrying that the heat produced will not be eliminated. As cooling units are expensive, investing in the most efficient systems upfront is vital.
- Server Space – Picking the most suitable server racks from the beginning will allow us to house the equipment correctly while leaving space for growth as well.
DATA CENTER PLANNING CONSIDERATIONS
Let’s review five critical design considerations and best practices when it comes to data center architecture.
1. Optimize Cooling
Having a good data center cooling design is one of the critical requirements. If our facility overheats, it could lead to catastrophic hardware failures that could cost millions to replace. Planning the cooling and airflow systems should be on our priority list. According to a Grand View Research report, 30% of a data center’s total energy expenditure is related to its cooling needs.
There are different types of air-cooling systems available. Here are the most popular:
- Traditional Air Conditioning – While industrial air conditioners are considered a reliable solution, they are energy-intensive.
- Water Cooling Units – Water cooling is more efficient compared to most other methods.
- Outdoor Air Cooling – This option is preferred if the data center is located in a region where the outdoor temperature is low for much of the year. It allows the use of outdoor air for cooling a data center.
- On-Rack Cooling – The back doors of an enclosure are replaced with a heat exchanger, allowing for precision cooling.
- In-Row Cooling – Through this strategy, a rack in the row is replaced by an air-handling unit, creating a shorter airflow path.
2. Smart Airflow Management
The next step to keeping our facility at the right temperature is managing the airflow.
A good airflow plan can decrease the cooling costs by up to 40%. We should consider these techniques when planning out our data center’s airflow strategy:
- Main Intake and Exhaust – Using hot and cold aisle containment has become standard for limiting cold and hot air from mixing. This method proved to offer excellent results.
- Server Rack Airflow – Good airflow within each server rack is essential, which can be achieved through blanking panels, for example.
- Segmented Aisles – The hot aisle/cold aisle strategy works best when using physical barriers placed above and around each aisle to direct the air.
3. Physical Security Plan
The physical security of our facility is essential to plan when designing a new data center.
Examples of physical security controls to ensure only approved individuals can gain access include the following:
- Physical Access Control with anti-tailgating/anti-pass-back turnstile gate
- Physical barriers preventing unauthorized access
- A single entry point into the facility
- Secured doors with biometric scanners
- Additional physical access restriction to private racks
- Closed-circuit television cameras (CCTVs) with full pan, tilt, and zoom features with video retention according to the organization’s policy
- 24×7 onsite security guards
- Network Operations Center (NOC) Services and technical team
4. Cable Management
Our cable management plan should focus on two main areas: the server racks and the cables.
To ensure an efficient installation, designing proper wiring schematics from the beginning is critical. Using a Data Center Infrastructure Management (DCIM) software to document the original design facilitates real-time transparency of physical network capacity throughout the entire facility.
5. Monitoring and Maintenance
According to a recent study, the Ponemon Institute reported the average incident time for a complete unplanned data center outage is about 2.5 hours, costing almost $700k.
It’s essential to monitor a data center environment correctly and take precautionary and proactive steps to reduce unscheduled equipment downtime. Data centers that don’t have a planned and preventative maintenance plan have an increased risk of asset failure. However, the task of managing data center infrastructure comes with its list of challenges.
According to the data presented in a White Paper, published by the EUDCA Technical committee, properly maintaining data center equipment can lower corrective maintenance costs by up to 90%,
A thorough maintenance plan for a Data Centre should include;
- Regular Inspections – In-house staff can regularly check the data center for any potential alarm alerts or inadequate efficiency use.
- Predictive Maintenance – The onsite employees can measure detailed data and monitor trends to forecast equipment closing to the end of life (EOL) and other needed repairs.
- Preventative Maintenance – We should conduct scheduled analyses to test, replace, clean and adjust the equipment, which helps us spot defects before something more severe occurs.
- Corrective Works – Quickly replacing defective parts and timely acquisition of end-of-life systems and equipment prevents impending failure.
Data centers come with large volumes of equipment that require continuous maintenance or reconfiguration. Thus, it is essential to have an easy and straightforward IT Maintenance Optimization schedule to track and manage all the systems. Given how expensive equipment is, it is paramount to use a system that catalogs all data center assets and tracks OEM and third-party maintenance contract data to find and optimize wasteful spending.