I recently ran across the Amazon S3 storage solution
> ) as an attractive
solution for online backup of data. Several products currently act as
local WebDAV adapters on top of the underlying S3 APIs.
These solutions are attractive because S3 does not support WebDAV
directly and the local solutions allow for local encryption and caching
(S3 charge for bandwidth). The most referenced solution is the non-open
source Jungle Disk (http://www.jungledisk.com/
> ). The homepage for Jungle Disk is worth a
glance for S3 pricing summaries ($0.15*GB*month storage + $0.20*GB
bandwidth). The open source s3dav project
> ) seems to have similiar aims;
although I haven't used it myself.
Jungle Disk itself seems relatively feature rich but immature. It caches
locally; supports S3-indepenent encryption keys; etc. Mapped drive
support on top of WebDAV is accomplished through generic third-party
products (e.g. http://www.webdrive.com
however, this integration is far from ideal since double caching, start
ordering dependencies, etc come up.
So what does all of this have to do with the NSLU2? The biggest
limitation (other than general immaturity) of Jungle Disk in my
environment is that the WebDAV server only listens on localhost (for
security apparently; wish they'd let me make that call). What I would
really like is an NSLU2 package that would expose a WebDAV listener to
my network where the contents were cached on the NSLU2 hard drive then
encrypted and synced with S3 in the background (essentially the
functionality JungleDisk is providing on my desktop today).
This would be far superior to running JungleDisk on each machine as the
cache would be shared across my network; file synchronization issues
(while background uploading to S3 takes place) would disappear for my
home network; and the startup ordering on issues would disappear since
the NSLU2 is always on (JungleDisk cannot run as a Windows service).
Of course with all of this the WebDav part might not even be required
(or could be optional) since the NSLU2 could bypass the webdrive-style
WebDav -> disk mapping software and just expose the sync area via the
normal SAMBA, etc sharing... however, I'm not sure if that's ideal as
WebDAV access protocols might offer more appropriate writing, locking,
etc semantics (given the necessary background sync with S3) to the
client OS... I don't know enough about WebDAV and this area to hazard a
So my first question is whether any similar solution already exists for
Amazon S3 or other similar service? If not then my question is whether
any existing NSLU2 packages might have a WebDAV listener, etc that seems
suited for rewiring behind the scenes to access S3 (in an implementation
that can perform reasonably on the NSLU2; I'm concerned that the s3dav
Java implemention might not be the best approach for the NSLU2; but I
haven't looked at it in any detail).
[Non-text portions of this message have been removed]