Researchers have discovered that a pinch of table salt can get more data onto spinning hard drive platters. A lot more: 6 TB in a single platter!

Considering the latest drives have up to 5 platters, that would be 30 TB in a single hard drive. Put 45 of those drives into a Backblaze Storage Pod and you would store 1.35 petabytes (!) in a single storage pod; more than 10 petabytes in a single rack.

Hello, Big Data. Meet your best friend, Salt.

A single Backblaze Storage Pod weighs nearly 150 lbs and takes two people to rack it. Each time Sean has gone to the data center to rack new pods (which happens every two weeks) he needed to bring another person for assistance. In the interest of giving Sean a hand, we bought him a friend. His name is Guido.

Guido is a $10k server lift and he works like a charm. He can lift the servers onto the very top rack and even has a special extender for the very bottom rack. Sean can now deploy Storage Pods without any help.

Meet Guido:

Sean working with Guido:

And a 3 second video of Guido doing the heavy lifting:

Ok, actually it was Chloe that built the first pod. And Greg that built the second. But those pods were built to store the geospatial information data for GINA – the Geographic Information Network of Alaska.

When searching for pics of Backblaze Storage Pods, I stumbled across Chloe’s Picasa album, and asked her why she needed boatloads of storage. Here’s her story:

The pod you stumbled upon in my Picasa album was one of two that I helped build for the Geographic Information Network of Alaska (GINA) in 2009/2010 when I was still a CS student. (I built the first pod, and finished up the build on the second, which was assembled by my colleague, Greg Wirth).

GINA is a small research group affiliated with the University of Alaska Fairbanks. We operate on grants mostly, so we have to be conservative and wise with our spending. That’s why my boss, Dayne Broderson, chose to use the open hardware that you so generously provide schematics to.

At GINA, we have many years worth of spatial and sattelite imagery, and we keep it all saved away for GIS specialists, cartographers, professors and students at the university to use. We also provide the majority of this data online through Web Mapping Services (WMS) on our website. It’s basically our own Google Earth, so millions of image tiles need to be stored. Given those circumstances, we needed something that would allow the data to be readily available for viewing. That’s where the pods came in.

Our pods are the first (that I’ve heard of, anyway) to be used as High Availability cluster nodes. That means if one pod goes down, the other pod takes over its IP and web services, without the users ever noticing. Since robustness is important to us at GINA, we also altered the design a little to allow for a RAID1 system partition. Otherwise, losing that one system disk could lead to unexpected downtime.

My colleague, Greg, made a mounting bracket that mounted something quite similar to this IcyDock inside our pods. It let us use two 2.5″ hard drives, which we then installed with RAID1 for the OS.
http://www.newegg.com/Product/Product.aspx?Item=N82E16817994087

Here’s the finished product:

The only issue we ran into when building these was the PSU. The Modu 87 was cheaper and more readily available than the other PSU listed on your site, so we tried the Modu first. Here’s a picture of the problem we encountered:
https://xanth.gina.alaska.edu/pod/IMG_8289.JPG

From there, we tested it out with some molex adapters, which helped with the connector angle problem, but didn’t hold up under testing. (I let out the Magic Smoke! >_>)

We finally got our hands on the Plexton PSUs with the custom cabling, and everything from then on was smooth. It cleared up a lot of space inside the cases too, since the cabling was much neater.

It went so well, in fact, and has been running so reliably since, that my boss is considering implementing the V2 design next. I’m kind of jealous that I wont be there to help them build it this time, since I moved away recently to study more high-availability clustering at LINBIT. But that pod build was one of the most fun and interesting projects I’ve done yet!

Thanks for inquiring about our pods,

Best regards,

Stefanie “Chloe” Edgar

Congratulations Chloe on building a high-availability cluster version!
And thank you for sharing the story!

It’s been over a year since Backblaze revealed the designs of our first generation (67 terabyte) storage pod. During that time, we’ve remained focused on our mission to provide an unlimited online backup service for $5 per month. To maintain profitability, we continue to avoid overpriced commercial solutions, and we now build the Backblaze Storage Pod 2.0: a 135-terabyte, 4U server for $7,384. It’s double the storage and twice the performance—at lower cost than the original.

In this post, we’ll share how to make a 2.0 storage pod, and you’re welcome to use the design. We’ll also share some of our secrets from the last three years of deploying more than 16 petabytes worth of Backblaze storage pods. As before, our hope is that others can benefit from this information and help us refine the pods. (Some of the enhancements are contributions from helpful kindred pod builders, so if you do improve your Backblaze pod farm, please balance the Karma and send us your suggestions!)

Quick Review – What makes a Backblaze Storage Pod

A Backblaze Storage Pod is a self-contained unit that puts storage online. It’s made up of a custom metal case with commodity hardware inside. You can find a parts list in Appendix A. You can also link to a power wiring diagram, see an exploded diagram of parts, and check out a half-assembled pod. The two most noteworthy factors are that the cost of the hard drives dominates the price of the overall pod and that the system is made entirely of commodity parts. For more background, read the original blog post. Now let’s talk about the changes.
Continue reading…

Ignite is a “Fast-paced, fun, thought-provoking, social, local, global” evening of “high-energy 5-minute talks by people who have an idea.” Specifically, as the presenter you get exactly 20 slides that auto-advance every 15 seconds. No running over. No slide remote control.

The theme of the evening was “Lean Startup” a set of principles (clearly communicated by Eric Ries) of building startups in a more efficient manner.

I presented the 5-step process of building Petabytes on a Budget. It was a blast and I highly recommend Ignite events and the Ignite format to anyone wanting to communicate an idea. As they tell you when you prep, a TV commercial tells an entire story in 30 seconds; if you’re crisp you can communicate a lot of ideas in five minutes.

Here is the presentation:

And here are the slides on SlideShare:

Thank you to O’Reilly for starting the Ignite concept and facilitating the community, Sarah Milstein for organizing the entire event, Tony Stubblebine for MC’ing, pariSoma for providing a great space, and DreamSimplicity for providing the video.

After we open sourced the Backblaze Storage Pod, hundreds of people contacted us expressing interest in building their own. A few have shared their experiences including the team at Vanderbilt University Institute of Image Science, where Hampton Albert and others have modified the original design for their own purposes. They explain their approach and use below.

Thank you and the Backblaze team for open-sourcing your hardware! At the Vanderbilt University Institute of Image Science, we support hundreds of researchers operating multiple high-field Magnetic Resonance Imagining (MRI), computed tomography (CT), positron emission tomography (PET), and ultrasound scanners. The institute’s MRI scanners alone generate over 200 GB of data per month. We need to reliably store these data for our research partners.

Without affordable storage solutions, we could not centralize storage and were forced to rely on optical backups for long term archive. Your “Petabytes on a budget” post was exactly the starting point that Hampton, our IT expert, needed in order to undertake construction of two 90 TB (raw) storage servers. As Hampton explains, hardware tweaking was a bit of an adventure, but we are now rolling out these boxes along with the open source clinical image and object management system DCM4CHE, the DCM4CHE enterprise-grade DICOM picture archiving and communication system, and open source Java Image Science Toolkit to provide secure, long-term data archive and rapid image processing.

With this affordable, redundant, and scalable storage system we will be able to provide indefinite data storage and retrieval for imaging studies without substantially increasing cost to our users.

Best Regards,
Bennett Landman, Ph.D.
Assistant Professor
   Department of Electrical Engineering (primary)
   Department of Biomedical Engineering
   Department of Radiology and Radiological Sciences
   Vanderbilt University, School of Engineering
Director of the Center for Computational Imaging
   Vanderbilt University Institute of Image Science

Amazing! Never did we think when we set out to build inexpensive cloud storage for our online backup service that the pieces would end up supporting MRI machines as archival storage.

Thank you Hampton, Professor Landman, and the rest of your team there for sharing this with us!

Twitter turned 5 years old yesterday. The service now has 200 million users (including @backblaze) who post over 1 billion tweets per day.

One interesting tidbit they recently mentioned was that all of those billions and billions of tweets compiled over the course of half a decade took a mere 20 terabytes of storage space!

In other words, you could archive all of the last 5 years of Twitter in just 1/3rd of an original Backblaze Storage Pod.

Happy Birthday Twitter! We wish you an amazing next five years!

As we approached the end of 2010, our vp of engineering sent an email:

Backblaze crossed the 10 petabyte mark in data storage for our customers.

While l see us constantly adding storage for our online backup service, it was one of those moments that made me think, “Wow, that’s a lot of storage.” Ten petabytes is roughly double the entire archive of the Internet.

We tried to visualize what 10 petabytes of storage looks like. Here are three takes on it:

#1: 10 Petabytes in Backblaze Storage Pods

#2: 10 Petabytes in Pixels by Drive Size Used

#3: 10 Petabytes of Drives Stacked Vertically (my favorite)

How would you visualize ten petabytes of storage?