Readers of this blog have seen previous commentary on Dropbox’s growth. (See: Dropbox Now Has 175 Million Users… and How Did Dropbox Scale To 175M Users?) It’s an impressive thing they’ve done. If you use Dropbox, you no doubt see its value. When I see that level of value and growth, I have to wonder how at-risk are they? Hence, you might also want to read And here’s how Dropbox loses 175M users…
- Test Fail Points: Dropbox forced that failure with hard reboots to live servers o to see of the failover really worked
- Keep Hardware consistent: by using the same hardware on all theri servers, they reduced “planning” time and, one presume, increased “action” time
- Use UTC: this allows all their servers to be in the same time zone and reduced the amount of processing/thinking that had to go into sychronization
- Release Often: keep ’em stable but keep ’em rolling.
A key point that Dropbox claims as central to their success was using the Python coding language. This is a bit outside my domain so I’ll have to take their word for it.
For the word on the street, click on over to Christina’s article: How Dropbox Scaled From 2,000 to 200 Million Users.