It’s arrived. Dropbox, the ubiquitous file-sharing platform, has finally submitted a rather light S-1, indicating an Initial Public Offering is on the way. From the various reports written so far, it’s likely the company is intending to raise around $500m.
The first thing that leaps out at me when looking into the announcement is the relatively low level of paid subscribers. Of the 500+ million registered users, only 11 million pay for the service (I’m one). As a result, I’m paying a subsidy on my fees for using the platform. Is that fair? The Dropbox model has always been “freemium” and of course even with 500 million users registered, we have no idea how many of those accounts are dormant or storing tiny amounts of data. We do know that there are 400 billion pieces of content, which gives an average of 800 files per account. Assuming paid accounts have more content, then the free ones could be storing very little data indeed. So perhaps I shouldn’t worry too much.
As a public company, surely Dropbox would like to convert those inactive users to paid ones? On a purely operational cost basis, I’d say that would be true. After all, it must cost less to service an account than the company charges an end user. However, in recent years, Dropbox has moved from an AWS operational model to building their own data centres in order to save money. What was operational cost got converted to capital cost, at least in the short term. Any increase in customer activity is going to mean more capital expenditure, rather than cranking up more VMs and S3 space in AWS.
Public vs Private
For a company the size of Dropbox, clearly, the move to private cloud was financially attractive. Economies of scale (eventually) made the transition worthwhile. With such a large amount of inactive data, there was clearly room for shifting a significant amount of content onto the cheapest platform possible. AWS storage costs have come down over recent years, but now are relatively static, whereas HDD capacities increase around 30% year on year and drive prices are falling. There’s a tipping point, based on capacity and activity where bringing that data back in-house makes more sense.
Dropbox appears to have created their own storage pods, in a similar vein to Backblaze, although there are claims that the density is much higher than Backblaze is achieving. When you have 500 petabytes of data to store, a custom solution is worth investing in. The idea of build versus buy is a perennial IT problem. From a software and applications perspective, building is more work, whereas buying off-the-shelf might not offer all the features required, especially with SaaS. Hardware has been more of a challenge and I would say that in storage, build is much rarer than buy.
However, if you have a single use for data (in this case sync ‘n’ share) then build could be an attractive process. This is especially true if a large portion of data is relatively inactive and the cheapest possible media can be used. Now here’s an interesting dilemma. If Dropbox was to add more functionality to their platform, how would that affect costs? Instead of operating a very large archive, the operational model would have to change to one of active data, with a totally different profile.
The Architect’s View®
Financially (at least according to this article in Hacker Noon), Dropbox is in a strong position. However, there are a lot of sync ‘n’ share solutions around and I don’t think the company has a great Enterprise penetration. At least not officially, anyway. With no direct cloud integration, is there a long-term future for Dropbox that could be challenged by the likes of AWS and Microsoft (who already have services) and of course Google? There are also services like Nasuni, CTERA and others that can offer similar file sharing for enterprise, plus platforms like HCP from Hitachi that can extend data assets to outside the data centre.
Dropbox as a name has become the de-facto file sharing solution, although it hasn’t quite become a verb in the sense of Hoover or Google. In terms of market share, the company may have already reached a tipping point where rivals can’t catch up, at least, not without buying customers at a huge loss. The IPO and next 12 months are going to be interesting times indeed.
- Why Dropbox decided to drop AWS and build its own infrastructure and network (TechCrunch, retrieved 26 February 2018)
- Dropbox mostly quits Amazon cloud, takes back 500 PB of data (Computerworld, retrieved 26 February 2018)
- Dropbox files S1 with $1B Revenue, Lots of Restricted Stock (Barrons, retrieved 26 February 2018)
- Data Mobility – Global/Scale-out Data Platforms
- Cloud as a Tier of Storage
- Is Google Drive Too Late to the Party?
- Object Storage Leading The Advancement of Software Defined Storage
Copyright (c) 2009-2022 – Post #1D83 – Brookend Ltd, first published on https://www.architecting.it/blog, do not reproduce without permission.