What is it like to build your own data centre?

Building a data centre that stands out so it can survive in a crowded market is tough, especially with the growing disparity between upstarts and giants, but nevertheless exciting. I speak from having worked on the ground, to eventually guiding the technology and business direction for our business — a cloud and data centre firm. It’s a never-ending process of optimizing and shrinking. A smart team oriented business setting with direction, is vital to squeezing innovation out from the engineering team. It sounds tense but that’s the game.

Demands hard work

First you need a team of skilled individuals that fit the entire stack to maintain and optimize the physical facilities, HVAC (heating, ventilation, and air conditioning), telecom, network, electrical, hardware, storage and server engineers. There is more. You also need to consider software and the licensing, recurring capital and labour costs, and the late nights in labs discussing continuous improvements.

Requires strict coordination

Much of daily life for these very hard-working engineers is driven by a few – but very important – key performance indicators that include uptime, capacity and performance at each layer of the stack. This is all being done on a shoestring budget, and our engineers take tremendous pride and ownership of ‘their’ systems. Budget constraints will always be there — unless you work for a web giant, this type of constraint culture isn’t apparent.

Power to steer

Depending on vendors for most of the stack severely limits the ability to compete and evolve. Being in control of the stack has many benefits, and controlling more of the stack allows DCs to decide their own destiny, this leads to more comfort and better SLAs that can be offered downstream to clients. This is why it’s no surprise that some of the well-known web giants [get others to] build their own custom hardware, lay their own fiber and telecom, and have multiple custom power and cooling solutions.

Getting in-house control of something like cooling isn’t a trivial affair, you have to consider humidity, air flow, saturation, pressure and PUE, and you have to have the right skill set to manage all this. It’s tough, but once you own this, you’re no longer beholden to an HVAC firm, you can keep costs low, and drive down prices.

Unless you have deep pockets to build out at scale and innovate with millions at hand, some of the other alternatives are innovating to be better or different, or being specific, and doing this on a shoestring.

This type of thinking is now necessary, and it directly affects what it’s like to build a DC.

A need to be unique

There are an abundance of tools that have enabled the creation and growth of software-defined data centres. By utilizing virtualization and hybrid clouds, multiple facilities are able to operate transparently as one. This is very important because it directly affects a DC’s cost structure and ultimately, the prices it sets and the profits it makes.

As you can imagine, it’s difficult to battle the web giants on their own turf. This is why business and market perspectives are more important than ever before. Having clean cabling or fast servers isn’t as important in the bigger picture. What you really want to focus on are things like carving out a market niche, efficiency, and unit margin per square foot.

Newer and innovative data centre builders are most focused on innovations that help lower capital and operating expenses. The type of data centre and how it is built must also be driven by market need and a specific offering. This means putting on a business hat and thinking about what’s a wash, cost substitutions and opportunity cost.

So, what is it like for us to build a data centre in this day and age of giants?

Well, it means staying calm and thinking about business fundamentals and questioning technology standards, and trying to do a lot with very little — this drives innovation. Yes, being specific in this way is risky, but the alternative of trying to do battle with giants who can pivot on a dime has only one unfortunate conclusion.

It’s an exciting time, and imposing constraints on oneself makes for creative thinking. It’s exciting and scary because the existing large firms are getting bigger and more entrenched. Money isn’t an immediate and existential issue for them, but it is (and should be) for an upstart. That means the time is ripe to start anew. We still sleep under our desks when we need to and argue over squeezing out watts from drive cycles, or aggregating virtual blocks over 100 miles. And, I imagine a lot of others in the mid-sized space do the same thing.

About the Author

Akshay Kalle is the CTO for the Pathway Group of Companies (pathcom.com), tasked with leading its infrastructure growth and innovation efforts. An “intrepreneur” at heart, he is a specialist in machine learning and big data and directs the firm’s innovation lab. A holder of multiple graduate degrees in computer science and mathematics, Akshay has carved out a name for himself in the area of applied machine learning. His accomplishments have included AI-driven email filters, smart OSS and monitoring systems, and predictive analytics tools used for finance and personal health, used by large carriers and international Fortune 500 firms. A second generation businessman, he aims to uphold Pathway’s brand virtues of ethical conduct, openness and relevance. When he’s not working, Akshay can be seen running, even when it’s -20 outside. Akshay can be reached at akshay.kalle [at] pathcom [dot]com.

Share this post on…