| Thread Previous • Date Previous • Date Next • Thread Next |
Yes, implementing a "Human" DCVS isn't easy. Dropbox, Apple's Time Machine, and several other projects have all /tried/ to to do so, but as far as I can tell they haven't succeeded. (It's gotten to the point where Dropbox decided that it just wasn't worth keeping file-revisions and now deletes revisions more than one month old. I haven't heard anyone complain, so I doubt that the feature was at all popular.) Both Dropbox and Time Machine remember small one-line edits that no one cares about. Its been impossible to create a "Human" GUI for viewing revisions because there are just too darn many revisions to be viewed.
I'm sure that Dropbox want to keep their user interfaces simple and intuitive, and this will be part of why their system works like this. There's an economic driver for this too, and I think this is very significant.
Dropbox wants to sell 50GB of file storage at $10 per month. That's a flat rate for a bunch of storage.
They will be banking on most of their customers using only a fraction of the full amount of available storage, because Dropbox is a cloud-based service using Amazon S3, and so they will pay for only what users use, not the full amount they are offering to users.
There's a problem with this approach if they are also offering to keep revisions indefinitely, or just remove revisions when the quota gets full. The problem is, over time, most users will be using the full 50GB. part for "live" data and the rest for historical revisions.
It costs Dropbox about the same amount to store a historical revision as it costs to store current revision of some files. But the value to the user is totally different. So, they will want keep only the most valuable revisions, and remove the rest. I guess that's why they've come up with this particular policy of removing older revisions automatically.
This will always be the case where companies are competing in advertising a quota of storage for a particular flat rate per month. They'll be hoping that, on average, users use a lot less than the full quota. And, they'll be developing software and policies around storing history that keep this average down.
It's a similar situation to broadband providers who compete based on advertising particular data rates, but make money based on the usage their customers actually make of the available bandwidth.
The alternative way to sell online storage is to charge based on the amount of data -- live data and historical revisions -- stored, and leave it up to users to decide how much history they want to pay for storing. That's the approach Amazon take if you use S3 directly. This is an attractive model for people who know exactly what they are doing, and are happy to have different bills each month. Most people just want online storage to work, and want to pay a predictable flat rate for a service that just works.
-- Steve
| Thread Previous • Date Previous • Date Next • Thread Next |
