Amazon Web Services is an unstoppable juggernaut in the cloud services world, contributing a projected $US7 billion to the retail giant’s revenues this year alone.
A big part of that dominance is thanks to Amazon S3, a super-popular cloud storage option that’s most easily described as like Dropbox for programmers to store all the data they need for web and mobile apps.
All of the images on Tumblr, for instance, are hosted with Amazon S3.
The killer feature of Amazon S3 is price. Because Amazon Web Services operates at such massive scales and tremendous levels of effieicncy, it’s able to offer storage at an ever-shrinking cost to customers. Right now, Amazon S3 pricing starts at three cents per gigabyte stored.
But there’s a new player in the market: Backblaze, a profitable, popular startup that lets users back up as much data as they want to the cloud for $US5 a month or $US50 a year.
Now, Backblaze is taking its eight years of experience with helping users back their computers up to the cloud and using it for “B2 Cloud Storage,” a new service that charges a measley $US0.005 per gigabyte stored — as much as 500 times cheaper than Amazon S3.
For price-conscious developers, that potential savings could make all the difference in the world.
And the really ironic part? If Amazon S3 had been cheaper eight years ago, Backblaze would never have developed the expertise it needed to compete in the first place.
No money, no choices
Backblaze’s founders met at an email security startup called MailFrontier, which was then snapped up by SonicWall in 2006, says CEO and co-founder Gleb Budman.
In 2007, one of the future Backblaze founders, still at SonicWall, was doing tech support for a relative who had accidentally and permanently deleted an important file. That engineer was shocked to find that the relative had absolutely no backups in place.
This led to a moment of inspiration, and Budman and other one-time MailFrontier executives were gathered to build a simple, straightforward way for people to back their stuff up to an online service for a predictbale, $US5/month price. In 2007, Backblaze was officially formed.
In order to stay disciplined on building the product, Budman says, Backblaze’s founders decided not to raise venture capital, instead relying on the team’s pooled savings to get the business off the ground. The thought was that taking venture cash would force Backblaze to focus on pleasing investors rather than building a sustainable business.
“At the most basic level, it makes people believe that money comes from VCs,” and not from actually making sales, says Budman.
In fact, they made the mutual arrangement to go without a salary for the first year of Backblaze’s existence — an arrangement that would actually extend to five years total.
Which was great, but it meant that Backblaze was starting from a disadvantage.
The original plan was to use Amazon S3 on the backend to build Backblaze as a simple storage system that would let them host their customer’s data.
There was just one problem: The pricetag. Amazon S3 was, and is, very cost-efficient for developers hosting a relatively modest amout of data.
For Backblaze, though, which planned on storing its customers’ entire hard drives in the cloud, Amazon S3’s price was too high. Backblaze would either have to charge its customers an unrealistically high pricetag, or else take a loss on their $US5/month pricing goal. Either way, Amazon S3 just wouldn’t work for them.
“Because we didn’t have any money, we didn’t have many choices,” Budman says.
“Really entertainingly terrible”
So Amazon S3’s pricing model was too expensive, and Backblaze didn’t really have the cash to buy commercial hardware from the likes of Dell or HP.
This meant that Backblaze had to get creative. In this case, creativity manifested itself as an ambitious plan to build its own high-capacity, low-cost storage servers, called “StoragePods.”
Backblaze hit upon a method for taking sixteen hard drives, stripped out of consumer-grade external hard drives by hand, and attaching them to a motherboard to make a simple server.
Those servers, with hard drives in tow, would then get placed into a custom server housing that they actually made out of wood. Those handcrafted, arsitinal wood-carved servers would then get driven out to a data center where they rented some space and installed.
A wooden server isn’t exactly industry standard, but it was enough to get Backblaze its first few customers. And, most importantly for the young company, these StoragePods were extremely cheap to manufacture.
Eventually, the StoragePod design evolved: It got a steeel casing, and they were able to cram in 25 drives with around 5 petabytes of storage each. And thanks to developing relationships with the suppliers of the casings and hard drives, costs stayed low compared to the revenue that the storage business was starting to bring in.
In the early days, of the company, Backblaze would actually have monthly StoragePod assembly parties, Budman says. They would go to their VP of Engineering’s one-bedroom apartment in San Francisco and just slap hard drives into cases all night long.
That was bad enough, but they’d also have to go through the tedious process of “burning in” those hard drives, Budman says, which involves running them constantly for a week straight to ensure they’re more reliable when they’re in the data center.
This meant that at any given time, there was always a server or two sitting on the TV stand, humming away day and night and heating the place up. Eventually, Backblaze shelled out for a shed in the backyard for burning-in purposes.
Those early days were “pretty entertainingly terrible,” Budman says.
But that work in cost reduction didn’t go unnoticed: When Backblaze released the blueprints for the StoragePod as a free download in 2009, over a million people took a look, which blew Budman away.
“This isn’t kittens or porn, it’s a storage server,” Budman says.
Competition with Amazon
The StoragePod has let BackBlaze ride to success.
The company is profitable, and was able to hold off on taking its $US5 million Series A round of funding until 2012, which was its fifth year of operation. Even then, Budman says, the funding was to fund growth, not a move of desperation.
Today, Backblaze operates its own data centres, and stores over 150 petabytes of customer data. It’s that focus on saving cash at all costs that gives Backblaze its competitive edge now, Budman says.
“Because we did that from the get-go, we have this storage that’s incredibly cost-efficient,” says Budman.
In other words, Amazon may have an obscenely huge reach, with gigantic data centres that means it can offer cloud storage at razor-thin margins, locking itself into the so-called “race to zero” with competitors like Google and Amazon.
But Backblaze does nothing but think about how to get the most bang for their buck, and squeeze as much storage capacity as it can out of its continually-evolving StoragePod design, Budman says. The newest Storage Pod, version 4.5, boasts 180 terabytes of capacity, and Backblaze still needs to put in 20 to 30 new ones per month (but no longer in anybody’s apartment, thankfully).
Which means that B2 Cloud Storage can offer even lower prices, and still make the company a healthy amount of money on volume.
And if Amazon S3 had just taken this stance when it started, Budman says, it would never be facing the threat of B2 Cloud Storage.
“The reality of it is, if this had existed at the time, we wouldn’t have had to build our own,” Budman says.
Disclosure: Jeff Bezos is an investor in Business Insider through hispersonal investment company Bezos Expeditions.
Business Insider Emails & Alerts
Site highlights each day to your inbox.