Need to move 2 petabytes from LA to San Francisco? If your data was already stored in a Blackbox, you could get over 350Gb/sec!
Here's the calculation:
By now I'm sure you've seen the news about Project Blackbox. I've been on the project for the last 6 months, and its one of the most exciting things I can remember. Rarely do you get to work on something that has the potential to change 50 years of standard practice. Ever since we started making computers we've had computer centers and then data centers, specialized rooms colocated with office employees. As others have now documented, this idea makes less and less sense as time has passed.
Do we have the perfect answer to this? Probably not yet, but the reason we decided to pre-announce this (it won't be in full production until next summer, though we will be working "hands-on" with customers soon) was to get the dialogue started so that when we ship the productized version (or, more likely, versions), we'll hit the mark for more real world situations.
While there's an incredible amount of innovation that's gone into the design, it's important to understand how critical the standard shipping container is to this concept. The shipping container we know and love today was invented in the 1950's by a man from N. Carolina named Malcom McLean. The first ship, namedIdeal-X, was a converted WWII transport vessel which sailed from the port of New Jersey to Houston.
There's two things which link this work to today's announcements. First is the economic discontinuity caused by the idea. Prior to the standard container it cost $5.86 per ton to load a ship. Afterwards the cost dropped to $0.16 per ton! This caused the head of the Longshoreman's union to say "I’d like to sink that s.o.b." at that launch of the Ideal-X.
I believe that over the next few years we'll see that the Blackbox concept will have as dramatic of an economic effect as Malcom's idea 50 years ago. People's way of thinking will dramatically shift - I've seen it in my customer visits so far - and they won't be able to think about datacenters the old way ever again.
The other bridge to the past wasn't about the container itself, but about what Malcom did next. In 1956, Malcom patented his container design. But instead of holding onto them and trying to sue anyone who copied the idea, he submitted the design to ISO and gave them a royalty free license. Today we'd call that englightened, back then it was just plain crazy, but it worked...today's there's over 18 million shipping containers in use worldwide. But Malcom must have known that challenges of having to compete in open market was well worth the trouble if the open standards made that market huge.
Malcom, for that I'm bestowing you with the posthumous title of Honorary Sun Fellow. Well done!
Wikipedia references:
On a family trip to Colorado this month, we made a short stop at the National Center for Atmospheric Research (NCAR) in Boulder. They run a nice lunchtime tour there, and it was interesting to hear what they're working on in the area of climate change and modeling. They bought a Connection Machine that I helped design way back in the Thinking Machines days, and I've always loved visiting there.
I couldn't resist asking one embarrassing question, though. When we were standing in front of their world-class supercomputing center talking about how it was helping analyze climate change, I had to raise my hand and ask how much CO2 that data center was responsible for? I know from our data center near Boulder that the electricity there is very "dirty" (over 2 pounds of CO2 per kWH, much higher than the national average which is around 1.3 pounds).
I got a graceful but data-free answer. I was hoping they were exploring alternate energy, but it sounded like they were mainly focused on lower power usage by the computers (which is tough with high end scientific computers).
For what its worth, my guess is that they're somewhere over 10,000 tons of CO2/yr (that'd equate to a 10MW facility), which is similar to driving a car 20 or 30 million miles.
One of the things I hear consistently in my job is that data centers are running out of some combination of space, cooling and/or power. Gartner and others have published figures saying that over half are up against at least one of these limits. So a pretty obvious question is, where is Sun at?
One of our folks took the initiative and gathered up some data. Looking at 10 data centers around the world, here's the numbers:
In addition, four were over 75% capacity on space, and two at over 75% capacity on power. In total, 6 out of 10 were max'ed out on something, and 2 more were over 75% on at least one thing. Only 2 data centers were "green" on all three.
Now that we we've got a baseline, I'll be tracking this here as things progress.