Response to: Economics of Carbonite/Dropbox — Flickerdown Dave Graham's Weblog

Response to: Economics of Carbonite/Dropbox

by dave on May 24, 2011

Image representing Dropbox as depicted in Crun...

Image via CrunchBase

This morning, Dave Cahill (@dcahill8) posted a rather in-depth look at Carbonite and provided a counter to their business model by including Dropbox.  While Dave is a conservative guy, I’m really not, so, I took the challenge to hopefully expound upon what I think is Dropbox’s COGS model.  The original article is here: The Economics of Carbonite and I’d STRONGLY advise you to read it prior to coming here. 😉

In my mind, there are two different scenarios:
a.) DevPay – in which each user is given their own Amazon identity but superimposed over their account is Dropbox’s metadata thus giving them a “slice” of revenue that Amazon makes off of data.
b.) Superuser – in which Dropbox simply provisions “buckets” for customers underneath their master account thus allowing them to tap into Amazon’s tiered pricing structure.

Scenario A makes SOME sense but I’m unsure if Dropbox can programmaticly sign up end-users for discrete S3 accounts without some sort of machinations with Amazon’s data and privacy policies. Consequently, what I’ll discuss is scenario B.

In scenario B, if a user signs up for trials, Dropbox provisions a “bucket” underneath their US Standard region S3 account and obfuscates the bucket name (randomizing using alphanumeric characters). By creating the bucket underneath their account, Dropbox can now actualize a cheaper scale than if they went with scenario A.

For a working example, let’s assume several things.
a.) Dropbox has 500TB of data stored in S3
b.) Dropbox is paying rates listed HERE: for US STANDARD region only (not US West which is more expensive)
c.) Dropbox uses Standard (4 replica; 2 DC) storage. I’ll provide metrics for RRS as well.

So, for 500TB of storage, Dropbox is paying:
storage (standard) per month: $0.095
storage (RRS) per month: $0.063
transfer (IN): $0.10
transfer (OUT): $0.08

Storage Total for 500TB (500,000 GB)
standard: $47,500
RRS: $31,500

transfer (IN): $50,000 (this # is skewed as they’re not transferring 500TB per month but this provides a wholistic ingest #)
transfer (OUT): $40,000 (this # is again skewed but would show what it’d cost to shove 500TB downstream from Amazon S3)

As you can tell, Dropbox is in a rather perfect position to cover their expenses without much effort and with minimal investment in infrastructure (I’m assuming they’re using some level of EC2 involvement for control plane, etc.) they’re on a perfect trajectory for profitability @ scale.

What do you think?



Enhanced by Zemanta
  • Great post and thanks for the link, glad to see people finally talking about the economics of Cloud.