Data Centers

Dropbox saved $75M by dumping the cloud, but your company won't

Dropbox saved money by building custom storage infrastructure, but before you get too cozy with those servers, consider this: You're not Dropbox.

Server huggers of the world, unite! According to the Dropbox S-1, the cloud storage company saved a whopping $75 million by shifting from public cloud storage "to our own lower cost, custom-built infrastructure in co-location facilities that we directly lease and operate."

If that sounds amazing, it is. It's also highly unlikely that you'll be able to get the same sort of cost savings because, well, you're not Dropbox.

No, really. You're not

I know, I know. It's tempting to think that you can run your infrastructure far more efficiently than Amazon Web Services (AWS), Microsoft, or Google. I mean, you know that Dell PowerEdge server inside and out, and can milk those CPU cycles like a pro. Never mind that you have a few thousand servers and they have millions. Or that their entire businesses depend upon running at a level of efficiency that your IT staff hasn't approached on even the most optimistic of PowerPoint pitches to the CIO.

You've got this!

Or...not. After all, not only are you not AWS, Microsoft, or Google; you're not even like Dropbox, which has more engineering talent than many small countries. It's not a legacy Fortune 500 enterprise—it's a Silicon Valley tech company that happens to sell storage.

SEE: Cloud migration decision tool (Tech Pro Research)

Actually, even that doesn't quite do Dropbox justice. As RetailMeNot software engineer Kyle Smith tweeted: "Dropbox runs their own infra[structure] using proprietary, vertically aligned tech engineered to guarantee data consistency upwards of ten 9s. They do this in a fierce commodity market where tenths of a penny matter and they still excel."

That is, Dropbox fails if it isn't amazingly proficient in storage. Moving from AWS S3 for storage to its own custom storage infrastructure helped Dropbox to double its gross margin, as Anaconda senior vice president Mathew Lodge pointed out. Lodge went on to note, "[W]hen you have an exabyte of stuff to store and it drives unit economics of your product, then it's worth it."

Most enterprises, including yours, don't have exabytes of data. And most enterprises don't run an entire business centered on storing data.

If you build it...

Even Dropbox, which found those unit economics to be so appealing in storage, continues to use AWS for its compute needs. Cisco's Jonathan Creasy is probably correct to argue that, "When you get to [a] certain point it is almost always better to invest in infrastructure." But for most companies, most of the time, they're simply not going to reach that point for any particular application.

Of course, with enterprises being what they are, infrastructure is always going to be a mishmash of new and old technologies. Even relatively new Dropbox, despite an early bet on public cloud, has never been fully invested in AWS. It couldn't be, as former Dropbox engineer Tom Cook advised: "Dropbox infra[structure] was always a mix of AWS and non-AWS, mostly because they adopted it in the very early days of AWS. Databases were on bare metal from the start, as EBS didn't even exist back then and EC2 perf[ormance]/reliability was bad for high IO MySQL."

SEE: Special report: The cloud v. data center decision (free PDF) (TechRepublic)

Your company, however, is likely nowhere near the level of investment in public cloud as Dropbox was, or is. Dropbox was forced to build what early AWS lacked, and chose to roll its own storage infrastructure when owning that part proved critical to the economics of its business.

Still, as Tom Krazit has written, "once certain startups turn into big companies with hundreds of millions of users, with computing needs that they've come to intimately understand, it can be far more efficient to set up computing infrastructure designed exactly with those needs in mind."

He's right, but his description almost certainly excludes you, which means you're likely to get far more value from investing in the public cloud than from persisting in private cloud dreams.

Also see

nocloud.jpg
Image: iStockphoto/Denis Isakov

About Matt Asay

Matt Asay is a veteran technology columnist who has written for CNET, ReadWrite, and other tech media. Asay has also held a variety of executive roles with leading mobile and big data software companies.

Editor's Picks

Free Newsletters, In your Inbox