DIY data centers: Backblaze reveals secrets of cloud storage

If Hollywood ever needs a cloud storage facility for a movie scene, they need to call the folks at online storage and backup company Backblaze. You can just picture Tom Cruise, playing a secret agent dressed in black, sneaking around the rows of shiny red pods in the white storage facility (half a city block long). This is what cloud storage should look like, and the company is now sharing their secrets.

You too can make a 135-terabyte, 4U server for $7,384.

Today Backblaze posted a new blog full of information most companies keep hidden about building and running a 15+ petabyte cloud storage farm.

“When we first open-sourced the storage pod design in 2009, it was the first time it had ever been done,” says Gleb Budman, co-founder and CEO of Backblaze, in an interview with VentureBeat. “500,000 people read the blog post and hundreds of companies around the world have since built the storage pods for their own purposes.”

Backblaze calls those 135-terabyte, 4U servers “storage pods.” They are self-contained units, composed of metal cases with commodity hardware inside, all designed to put storage online. The company shares images of a half-assembled pod. The cost of the hard drives dominates the price of the overall pod and the system is made entirely of commodity parts. For more background, read the original blog post.

Today’s Backblaze post includes detais on how to make a version 2.0 storage pod, data on the total cost of ownership, the impact of heat on drives and more. This is one of the longest blog posts you will read today. Even if you aren’t interested in building a cloud farm, Budman’s writing is extremely novice-friendly. It provides some interesting insight to a technology most of us use — unconsciously — but few will ever see or think about.

“You’re welcome to use the design” is a phrase rarely uttered these days in Silicon Valley. Budman has it right there in his blog, in writing.

“We benefitted greatly from many other open-source projects and the efforts that thousands of people put into them,” he explains to VentureBeat. “We wanted to share some of that good karma back.”

This openness is unusual for a hardware company. Imagine Apple inviting people to fully examine and contribute to its products.

“We see ourselves as a software/service company at heart, trying to build the easiest online backup service available,” Budman says. “We built the hardware out of necessity because we could not find anyone selling the hardware we needed. Most of the storage servers available at the time cost more than $1,000 per terabyte at a time when drives were less than $100 per terabyte. We see our secret sauce as the backup technology on the client and software technology for the cloud.”

Budman says he gets feedback from “helpful kindred pod builders,” including Justin Stottlemyer from digital photo storage and sharing company Shutterfly and another unidentified person who setup the Google Group “OpenStoragePod.” Budman says hundreds of individuals have emailed ideas and feedback to Backblaze over the last year and a half. Having this much conversation about a piece of hardware is unique in the competitive cloud industry.

With data center space (and power) at a premium, minimizing storage cost means maximizing density. Backblaze’s new pods store twice as much data in the same space as the old ones. One datacenter rack containing 10 pods costs Backblaze about $2,100 per month to operate (the price is divided into thirds for physical space rental, bandwidth and electricity). Doubling the density saves the company half of the money spent on both physical space and electricity. The new pod cabinets squeeze one petabyte into three-quarters of a single cabinet for $56,696.

So how does Backblaze make money? The company’s business is to provide an unlimited online backup service for about $4 a month (if you buy for two years.) There aren’t any hidden caps on storage and Backblaze automatically backup all data, including on external drives.

Backblaze is going against the Silicon Valley grain in another way: The company is employee owned. It has not taken any venture funding.

“If we raised VC funding, we could theoretically burn money to start and hope to figure it out later,” says Budman. “Since we decided to bootstrap the company, we needed to ensure the business became cash-flow positive quickly. That forced us to innovate on the cost of storage.”

The company is now profitable and cash-flow positive, Budman says.

Budman says that to stay profitable, they have to keep costs and expenses low. He says that one of the hidden datacenter costs is the salary of employees who deploy pods, maintain them, replace bad drives with good, and generally manage the facility.

“Backblaze has 16 petabytes and growing, and we employ one guy whose fulltime job is to maintain our fleet of 201 pods, which hold 9,045 drives,” says Budman.

That guy is Sean Harris, and once every two weeks he deploys six pods during an eight-hour work day. Another employee (sometimes it is the CEO) helps Sean lift the 143-pound pods into place. He does the rest. Sean also spends one day each week replacing drives that have gone bad. As of this week, Backblaze has more than 9,000 hard drives in the datacenter, the oldest was purchased four years ago.

“I’ve worn a lot of hats in the tech world starting off in technical support, working in the field as a data warehouse consultant, DBA, and QA Engineer,” says Harris in an email to VentureBeat. “So working in the heart of the operations is both challenging and fun. But my Christmas wish to the company is another IT person… or two.”

All told, Sean Harris replaces approximately 10 drives per week.

In case you were curious, the company’s favorite drive is the Hitachi 3TB drive (Hitachi Deskstar 5K3000 HDS5C3030ALA630) because of its low power demand and reliability. Their least favorite is the Western Digital WD20EADS  and the Seagate ST32000542AS.

Now go ahead and build your own storage pod facility. Please send us a picture when it’s ready!

1 comments