SpiderOak Review

Dan Mayer asked about my experience with SpiderOak backup. I realized if I was going to type up my thoughts I might as well post them where others could benefit from them as well.

SpiderOak is a cloud-based backup, file synchronization, and sharing service. Unlike services such as Dropbox and ZumoDrive, the emphasis is on the backup part rather than the sync-and-share part.

At present, I have about 384GB from four different machines backed up on SpiderOak’s servers. After SpiderOak’s de-duplication and compression, this comes to about 115GB of usage, so I fit easily in the 200GB plan.

Since I started using SpiderOak  a year or so ago, there have been several instances where I was able to retrieve an accidentally deleted file, or grab a set files that I had only saved on another machine. I’ve also switched over to using SpiderOak for synchronizing the large set of files that I keep consistent across all my machines. I used to use a VCS for this, but that was tedious and breakage prone. With SpiderOak it pretty much just works. I’ve also made occasional use of the filesharing features. So it has definitely been useful to me.


  • Unlike many cloud-based storage systems, SpiderOak uses zero-knowledge encryption. That means that data is chunked and encrypted on the client side, and all the SpiderOak servers know is opaque data. For me, this is an essential requirement for a full-fledged backup service (as compared to a sync-and-share tool).
  • They support and even encourage you to use it for as many machines as you want. There is no cap on number of systems backed up.
  • It works on Windows, Mac OSX, and Linux. This was a big selling point for me, since I tend to have heterogeneous systems, or at least I did for a while.
  • The de-duplication works well, and means that (for instance) if you have your entire music collection on two separate machines, the backup won’t take any extra space for the second machine. And then you can have SpiderOak sync the two collections!
  • SpiderOak keeps version histories of files, not just a single backup.
  • At $1/GB/year, it is cheaper than many of the alternatives.


  • Because it is zero-knowledge, SpiderOak can’t delegate any of its processing to the cloud. All the chunking, deduplication, encryption, and sync calculations have to be done on the client side. This can put a considerable burden on your systems in terms of CPU and memory usage. It’s not unusual for CPU and disk usage to suddenly spike for a few seconds after changing some files, as SpiderOak detects the change and backs it up.
  • In particular, it has an issue when its index contains backups with an exceptionally large number of directories – as happens when backing up the full contents of linux systems. In this case, its ambient RAM usage can balloon. It’s not unusual for SpiderOak to consume 500MB-1Gb of RAM on my machine. The SpiderOak developers are working on moving all of the file index into an on-disk instead of in-memory database, at which point this should no longer be an issue. I don’t know what the ETA on that is, though; I’ve been waiting for a while.
  • Again probably as a result of its zero-knowledge nature, it isn’t as snappy a file-syncing tool as something like Dropbox, nor does it have options for having shared sync folders across multiple users. So there are somethings that I still use Dropbox for.
  • The UI is only so-so, and as yet doesn’t lend itself very well to managing your backup set. I’d love to do soem selective pruning in my backup fileset, but there’s no easy way to zero in on which directories are taking up the most space.
  • SpiderOak does not provide any kind of hardcopy backup service. I’d happily pay them extra to send me a stack of DVDs with the opaque, encrypted, de-duplicated data on them once a year – preferably along with a boot disk that had SO preinstalled –  but no such luck.
  • I can’t find any detailed technical information on SpiderOak’s cloud. I’d love to know more about their physical security and how much redundancy they have.