[tahoe-dev] General questions about TAHOE-LAFS, local filesystems and P2P.

[Martin Baldan]

Hello,

I'm basically just a curious potential end user of this awesome technology.

I've read a couple of posts that got me thinking:

[Freedombox-discuss] Tahoe-LAFS is not a filesystem
http://lists.alioth.debian.org/pipermail/freedombox-discuss/2011-November/003162.html

[tahoe-dev] "BitTorrent for storage" is a bad idea -- [p2p-hackers]
P2P file storage systems
https://tahoe-lafs.org/pipermail/tahoe-dev/2011-February/006150.html


This is a little example of what I think a nice user experience could
be like, realistically (ie, assuming files and applications as
currently understood are still relevant):

I create a document, let's say a LibreOffice presentation. I save it.
I modify it and save it again. I right-click on it (or similar
procedure) and I can see that a copy is stored locally, another copy
is stored in an online server I pay for (fully encrypted), and a third
copy is stored in my laptop (or smartphone). They are all synced,
unobstrusively (I don't experience any latency, syncing just works in
the background).


Now, let's say I want to share this document with a friend. I open a
menu on the file (right click or whatever), select "share", select
friend (like in Google docs). Now he will have a notification that I
shared a document with him.  My friend right-clicks on this document
and he selects "follow this document". Now If I delete my document
later, my friend will still be able to see all the snapshots up to the
time I deleted it.

My friend likes the document, so he recommends that I publish it. So I
simply share it with a some popular swarm of peers.


My question is, where does Tahoe fit in this picture, if at all? I
mean, what would be the best way to merge local storage, single-user
distributed storage, F2F (like Tahoe seems to be) and full-scale P2P,
at the lowest level possible and in a transparent, but configurable,
way? Since Tahoe is not intended to fill all those roles, how could it
be combined with other services in order to produce a seamless user
experience? Should Tahoe be the manager and delegate to other
filesystems, should it be the other way around, should they coexist
and be managed by some upper layer?

And how does Tahoe relate right now with other projects such as
XtreemFS (very similar to Tahoe) , Camlistore (more emphasis on
metadata, less on duplication logistics), OneSwarm (privacy-preserving
P2P), Tribler (a trackerless Bittorrent), to name a few? I would say
XtreemFS is a direct competitor (but there could also be cooperation).
As for the others, maybe they could be somehow combined with Tahoe to
build an awesome user experience. What do you think?

Best,

 -Martin