Whenever I speak to audiences about the latest IT advances, the topic of storage always registered a ho-hum-whatever on the let’s-get-excited-about-technology-o-meter — but now it’s just about the coolest thing out there.
Let’s start by taking a quick look at how the relative costs of storing a gigabyte of data have come down over the past few years:
- 1980: $3,000,000
- 2000: $30
- 2011: $0.035
The reason for this dramatic cost decline is the continuous improvement in disk density — i.e. how much data you can cram into a square inch of media.
Today, we are at an all-time high of 1 terabyte p/in2 — and scientists in several of our partner companies are working on technologies that, over the next 4 to 6 years (commercially available), could increase that density to multiple terabytes p/in2.
Meanwhile, last August Harvard scientists announced they had been able to store 700 terabytes into a single gram of DNA!
Now when I started computing (many years ago) 700 terabytes would have consumed every square foot of floor space in the Sabre campus — but with this advancement you could comfortably walk around with it under your eyelid!
Sadly, this does not mean our data centre costs will suddenly start to fall — because (and this is great news for consumers) instead we will all be able to leverage these advances to do so much more than we can today within the budgets we currently have.
Over the past four weeks, I have travelled to our main global development centres in the U.S., Poland and India to talk about our technology strategy. In these discussions, we always talk about where technology has come from and what our technology partners are doing now. This approach gives us context on where technology is headed so we can start to plan for its eventual arrival now.
Advances in storage are obviously going to play a big role in our technology direction. It’s not just about costs, either. Low-cost, high-performance, non-volatile memory will significantly improve customer experience, the way our applications work and the technologies used to make them work. But more on this in my next blog.