Data centers have their earliest roots in the huge computer rooms of the 1940s. A little over half a century, they have come a long way—helping companies like Google and Microsoft change the face of technology. From transforming how we commute (hint: Uber) to how we travel (hint: Airbnb), from aiding the arrests of most-wanted criminals to providing real-time data on the go, data centers have played an irreplaceable role in how we process and consume data today.
That being said, it is not uncommon for data centers to get flak once in a while, whether it is for their controversial consumption of energy or questions on the data capacity. In this blog, we will discuss data centers’ capacity and their capacity planning to help you choose and optimize the best magnitude for your business needs.
Storage Capacity of Data Centers
According to Wikipedia, one room of a building, one or more floors, or an entire building, can hold 1,000 or more servers. However, a data center’s capacity depends on various variables such as its size, efficiency, built, and technology being used. Data centers continue to grow stronger than ever, with over 500,000 facilities worldwide and an estimated capacity of more than 1500 exabytes (1 exabyte is roughly equivalent to 1 billion gigabytes).
Still, it is hard to ascertain the exact capacity of data centers, as companies categorize this kind of information as confidential and hardly make it available for public consumption. To give you a picture, a Gartner report from 2016 predicts that Google, which inarguably has the highest data center capacity to date, had 2.5 million servers at that time. Similarly, China’s Range international Information Group is the world’s largest data center with 6.3 million square feet of space. Moreover, every day we can hear news of companies investing billions of dollars to set up new data centers, revamp old ones, and make them as sustainable as possible.
Is it Enough?
As the amount of data being generated continues to grow tremendously, the tech world has often experienced periodic upsurges of panic about whether or not the world’s data centers have the storage capacity to manage all that information. But with over 2.5 quintillion bytes of data being created every single day, will it be enough?
After all, the tech space is littered with thousands of examples of the complications associated with the data center strategy mistakes around performance and capacity. For instance, Lady Gaga’s fans brought down the vast server resources of Amazon after her album “Born This Way” was offered online for just 99 cents. These kinds of occurrences are not uncommon, even for the big guns of the industry equipped with the best infrastructures and technologies.
These examples are reminders of why data center managers have to make sure their data center strategies are ahead of the organization’s scaling needs. It’s also critical to watch out for unprecedented peak requirements that can overwhelm your current center. The only way to do so is through data center capacity planning.
Data Center Capacity Planning
Data center capacity planning refers to the process of planning for current and future software, hardware, and other data center infrastructure requirements within a given time period. It can also be referred to as a form of IT capacity planning that reviews and analyzes current data center usage to plan for data center capacity expansion, contraction, both or none. Here are some tips to help you plan your data center’s capacity:
Correct Assessment of Hardware
Assessing current and upcoming hardware capabilities fall under two divisions—inventory and performance. It’s tempting to include every new advanced device or technology, but knowing exactly what you have will enable you to understand exactly what your data center is capable of (and not), in addition to determining its future capabilities.
For an inventory, you have to document each server’s name, make and model, OS, memory and disk space, number of CPUs, and network interface cards. On the performance side, make sure to document each CPU queue, memory paging, disk inputs/outputs, actual network speed, and a percentage per CPU used.
Strong DCIM is Vital
A data center without data center infrastructure management (DCIM) can be compared to a computer without an operating system. One of DCIM’s main functions is the culmination and analysis of metrics throughout the data center, including power usage, environmental conditions, and status of equipment. On a daily basis, it helps data center staff handle operational efficiency by identifying potential sources of waste. DCIM also helps to identify how the inclusion of new servers will change temperature and power consumption. All these insights can help improve capacity planning effectively.
Addressing Future Needs
Irrespective of what option works the best for you currently, it is important to make plans for what happens when you need to scale up. The costs to run a traditional data center is not cheap at all, and you’ll likely need to consider—regardless of the system you are using now—the possibilities offered by the four types of data centers:
- Scientific computing systems operated by national laboratories
- In-house facilities owned and managed by the company using them
- Public cloud providers like Amazon and Google
- Co-location systems where servers are placed together for utilization my multiple clients with their private cloud storage system
Dissolve silos between business stakeholders and IT
Most of the time, communication between data center workers and business stakeholders is sparse. Industry expert Bill Kleyman claims that one of the best ways to succeed at capacity planning is to close this gap. It is important that upper-level IT managers align their data center forecasting views with the goals of the business. It eventually helps them better forecast growth within the data centers. Likewise, lack of coherence between facilities (those who are responsible for power consumption and environmental monitoring) and IT (them that hands racks, cabinets, servers, and other equipment) will hamper a data center’s ability to respond to changes in overall capacity.