Dropbox saved $75M over two years by building its own infrastructure (2018)

#1

Post by simonebrunozzi » Wed, Nov 18, 2020, 5:36 PM UTC

Dropbox saved $75M over two years by building its own infrastructure (2018)
geekwire.com

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#2

Post by shay_ker » Wed, Nov 18, 2020, 6:14 PM UTC

Does anyone know what the break-even point is for building/managing your own infrastructure, vs. using cloud services?

I'm also curious - what's the rough time/cost of moving to your own infrastructure? I'm sure it depends on lots of things, but looking for a ballpark. 1 year + 10 engineers? 5 years + 100 engineers?

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#3

Post by calderarrow » Wed, Nov 18, 2020, 6:17 PM UTC

I think this quote really hits the nail on the head and confirms what a lot of people may have intuitively known about the value of cloud providers:

  But 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.

I think the main advantage of cloud providers is to offset the risk of purchasing equipment that eventually is no longer needed, which is ideal for younger companies that are still trying to reach their market capacity or unsure about whether they'll still be around in a year. Of all the things on a startup's todo list, I can't imagine setting up their own infrastructure is the best way to improve profits or revenue.

But once the constraints of a userbase are more established, it should be easier to migrate off these platforms, since their pricing is optimized for users of all business sizes and use cases, whereas your specific hardware can be optimized for your specific users.

My biggest question is whether cloud providers could achieve a scale where they are able to offer the most optimal infrastructure costs for specific businesses. Maybe this is the case for smaller or mid-size companies, but I'd be interested to see where the inflection point lies.

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#4

Post by gravypod » Wed, Nov 18, 2020, 6:18 PM UTC

shay_ker wrote: post #2

Does anyone know what the break-even point is for building/managing your own infrastructure, vs. using cloud services? I'm also curious - what's the rough time/cost of moving to your own infrastructure? I'm sure it depends on lots of things, but looking for a ballpark. 1 year + 10 engineers? 5 years + 100 engineers?

Depends on what your use case is. For GPU compute, in my experience, it's pretty much 2 months of continuous usage.

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#5

Post by hn_throwaway_99 » Wed, Nov 18, 2020, 6:19 PM UTC

shay_ker wrote: post #2

Does anyone know what the break-even point is for building/managing your own infrastructure, vs. using cloud services? I'm also curious - what's the rough time/cost of moving to your own infrastructure? I'm sure it depends on lots of things, but looking for a ballpark. 1 year + 10 engineers? 5 years + 100 engineers?

I mean, the more important question for me is, is running your own infrastructure something that should be a core competency. For Dropbox, the answer seems like an obvious yes. I mean, virtually their entire product is around data storage. I would be shocked if they COULDN'T save money to build a storage system tailored to their needs.

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#6

Post by ebiester » Wed, Nov 18, 2020, 6:21 PM UTC

Just as a reminder: You are not Dropbox. You personally need to run the numbers with both scenarios. Depending on your use case, it may be cheaper to be in AWS, and it may be cheaper to be co-located. If you are as big as Dropbox, it may be cheaper to build your own data center.

It reminds me of the arguments for and against K8s. Most of the discussion is based on use case, and not considering the solution space for all solutions. E.g. if I'm running an Erlang solution in K8s, or have a small number of servers, I may have better options.

It isn't only about scale. It takes understanding your system and your needs in depth, and any short cuts lead to cost inefficiencies.

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#7

Post by macspoofing » Wed, Nov 18, 2020, 6:21 PM UTC

Not very surprising. Cloud infrastructure is flexible, but very expensive - certainly much more expensive than hosting your own servers.

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#8

Post by sharadov » Wed, Nov 18, 2020, 6:22 PM UTC

I work for a company that does about 100M revenue per year. Run everything on prem, we did the numbers if we moved to AWS, it was almost 10 X of our current spend. Again , depends on your business, ours is complex and compute intensive. But at scale, on-prem always wins.

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#9

Post by rossmohax » Wed, Nov 18, 2020, 6:22 PM UTC

I remember Gitlab blogged that they are going to use baremetal instead of Azure with some numbers showing benefit of doing so. AFAIK they quickly reversed decision and didn't even try to implement that plan, moved to GCP instead.

Re: Dropbox saved $75M over two years by building its own infrastructure (2018)

#10

Post by ebiester » Wed, Nov 18, 2020, 6:22 PM UTC

shay_ker wrote: post #2

Does anyone know what the break-even point is for building/managing your own infrastructure, vs. using cloud services? I'm also curious - what's the rough time/cost of moving to your own infrastructure? I'm sure it depends on lots of things, but looking for a ballpark. 1 year + 10 engineers? 5 years + 100 engineers?

There are no shortcuts here. You really have to understand your use cases.