How To Build A Modern Data Centre?

Data is everywhere. By 2025, the amount of data generated is predicted to cross a staggering 175 zettabytes. The data explosion has amplified the importance of data centres. Technology giants like IBM and Microsoft are aggressively investing in setting up data centres globally. India alone saw 14 major data centre investments in 2020 (till September).

Data centres, often referred to as the backbone of the internet, are used to store, communicate, and transport the information generated on a daily basis. 

Given the kind of heavy lifting they do in the modern computing era, data centres need to be highly efficient, robust, and up-to-date. Besides, a good modern data centre has to be cost-effective, energy-efficient, and responsive. So the looming question is whether the data centres today are efficient enough to keep up with the accelerating demand?

A Forrester report on the modernisation of data centres made a few interesting points:

• The report found data centre modernisation, especially for midmarket firms, is not good enough to meet organisational needs.
• There is a general lack of agility
• Greater investment in data centre modernisation is directly linked to better business outcomes.

Below, we look at strategies to build a modern data centre.

Adoption Of Hybrid Data Centre

A hybrid data centre has mixed computing and storage environments. It combines the benefits of on-premise data centre and private cloud services (private or public). A hybrid data centre quickly responds to changing business needs and ensures all the workloads are handled cost-effectively.

Better Data Centre Infrastructure Management (DCIM)

For running a data centre, it is important to have real-time and accurate measurement of the data centre environment. A DCIM software integrates the monitoring of the entire IT infrastructure into a centralised system. It offers information on how the data centre environment is configured, the resources assigned to physical or virtual services, and even temperature, humidity, and airflow measurements. It also offers improved uptime, greater energy efficiency, and mitigates human error. 

Modularisation

Traditional data centres are labour-intensive and need huge investments for upscaling. A possible solution is the modularisation of data centres. The entire data centre is compartmentalised, and each module is designed separately for scaling, configuration, and power load according to a unified standard.

Modules act as building blocks and can be added or removed from infrastructure, allowing the flexibility of having resources on-demand. Plus, over-provisioning can be avoided. Though modules are not scalable on-demand for most cases, they offer interoperability, streamline the infrastructure, and reduce administrative burden.

Software-Defined Data Centre

A software-defined data centre (SDDC) pools infrastructure resources, standardises management tools and enables policy-driven provisioning. SDDC is increasingly being seen as a means to reduce costs and help in application modernisation.

SDDC integrates virtualisation layers and creates a single hyperconverged environment. Having a single managed environment standardises management across virtualisation layers.

Making Data Centres Energy Efficient

In 2017, US-based data centres alone utilised over 90 billion kilowatt-hours of electricity. As per another report, data centres globally consumed 1 percent of all electricity consumed in 2018. With an increasing number of data centres, energy consumption is bound to go up.

Companies can adopt the following strategies to reduce power consumption:

• Since a great amount of power is consumed for cooling and air conditioning to protect the equipment from overheating, it is important to optimise cooling operation.
• Synchronise server capacity and load using monitoring and management tools
• Due to negligence or plain oversight, certain assets are forgotten, turning them into zombie servers. These servers remain powered and continue consuming energy. Such assets must be identified and removed.