The Manufactured Assembly Line Data Center
Why is an individual computer server different from a data center, other than scale? After all, the physical characteristics of a server are very similar to those of a data center.
Let's Compare the Characteristics of a Server to those of a Data Center
Server Housing (case) = Data Center Building
Case Open Alarm = Data Center Entrance Security & Environment Management Alarms
Fans = Air Conditioning
Power Supplies = Power Supply (transformers, UPS, distribution, PDUs)
Alerts/Alarming = Data Center Monitoring and Alerting
Administrator = Data Center Staff/Manager
Why is the above comparison of a server and a data center important?
"We need to move away from building custom data centers". There I said it, man that hurt. As a data center guy, I can't stand the idea that there will soon be a time when building a unique facility for my company will most likely be the wrong thing to do. Before you start throwing things at your computer screen and yelling my name in anger, consider the following historical examples of the automobile and the personal computer:
The automobile - Prior to the invention of the assembly line by Ransom Olds cars were handmade and individually assembled.
A Little Assembly Line History:
In order to keep up with the increasing demand for those newfangled contraptions, horseless carriages, Ransom E. Olds created the assembly line in 1901. The new approach to putting together automobiles enabled him to more than quadruple his factory's output, from 425 cars in 1901 to 2,500 in 1902.
Olds should have become known as "The father of automotive assembly line," although many people think that it was Henry Ford who invented the assembly line. What Ford did do was to improve upon Olds's idea by installing conveyor belts. That cut the time of manufacturing a Model T from a day and a half to a mere ninety minutes. Henry Ford should been called "The father of automotive mass production."
One could also argue that since the advent of the assembly line and mass production, there continue to be innovations that reduce the complexity and "individual" nature of many automotive components. Most automobile manufacturers don't build their own windshields or individual parts anymore. Instead they are mass produced by others who can apply standards across multiple car lines. You can still buy a handmade car, but you definitely pay the price.
Personal Computers (PCs) - Prior to 1982 many of us bought components and made our own PCs. Who would build their own server or PC today? Maybe high end gamers or organizations looking to solve a very unique problem whose solution is more important than the cost. Generally speaking, building your own computer today would be much riskier, and costlier than buying one pre-built.
Increased attention and new money will accelerate the push to a standard build data center model.
Historically speaking most of us in the data center space understand that there's still room for unique design as we push the limits towards the best, most efficient and flexible data center. However, the data center has gained so much visibility over the last five years that it is no longer the black box that it used to be. This public awareness has dramatically increased investment dollars in the data center industry and created a large number of data center experts. The investment and influx of experts mean that change and improvements happen at a much faster pace than in the past.
If manufacturing automation and standards can be applied to cars and PCs why shouldn't they be applied to Data Centers?
So, I guess what I'm saying is that the data center is ripe for the picking, just like the manufacture of PCs was in the early 80s. As we get to the point of diminishing returns relative to the efficiency and modularity of data centers, companies that continue to try and build their own will be at a distinct disadvantage. We all know that time is one of the most costly of our resources. If you can implement new capacity in a matter of weeks vs. a matter of years, you're better positioning your enterprise to leverage their IT investments to meet changing business demands. You may believe that you could build a data center that's better than pre-fab modular boxes, and you could be right. However, just like in the case of the PC you'll likely discover that your risk and months of lost hours don't justify the small expected improvement in efficiency.
Who is going to win in the modular data center space?
What am I a psychic? I don't have a clue, but the writing is on the wall, first it was inflexible containers and infrastructure that wasn't ready. Now the containers have improved and we've got pre-fab modular designs available and IT infrastructure is quickly catching up. The key driver towards moving to a low cost standard "capacity of compute" model for your DC space will be when the majority of our applications and infrastructure can be distributed and portable through a combination of cloud technologies.
Hold on tight it's going to be a fun and probably rocky ride, but in the end businesses should win and that's what's important. Who would you name as the Ransom Olds or Henry Ford of the Data Center?
This link is to a related blog I wrote a little over a year ago: