[Tahoe-dev] my reply to wmf

zooko at zooko.com zooko at zooko.com
Sun May 6 23:53:16 UTC 2007


edited

------- Forwarded Message

To: Wes Felter <wesley at felter.org>
Subject: Re: Tahoe! :-) 
From: zooko at zooko.com
Date: Sun, 06 May 2007 17:17:29 -0600


> Hey Zooko, I saw the announcement and looked into it a bit, but as a  
> p2p-critic rather than a Tahoe developer I didn't see anything new  
> there compared to the good old Mojo Nation days. I am a little  
> disappointed that Foolscap is just using SSL rather than end-to-end  
> encryption. More worrying is the lack of any mention of NAT  
> traversal. So I guess I'm glad you have such well-loved, tested, and  
> documented code.

You're right about all that.  Could I quote you on the tahoe-dev list or a
private Cc: to the core developers?  Perhaps we need to write more docs about
intended future development.  We're doing an incremental-improvement scheme
which I'm very pleased with, but it also means that our plans to fix some of
these problems before the first Tahoe-based product aren't apparent.  You can
find clues about these plans in roadmap.txt [1], and denver.txt [2].

I really liked your observations about storage bandwidth and FEC bandwidth.
Could I forward your entire message to the tahoe-dev list?

So yeah, anyway, what I'm excited about with Tahoe is not what it already does,
but that the development team, the leadership, and the extant code base are all
things that make me happy.

Regards,

Zooko

[1] http://allmydata.org/trac/tahoe/browser/roadmap.txt
[2] http://allmydata.org/trac/tahoe/browser/docs/denver.txt



> Given that the asymmetric bandwidth problem and the FEC expansion  
> problem compound each other, a FEC service run by high-bandwidth  
> peers would seem to make sense. A peer could upload one copy of data  
> (pre-FEC) and then perform a verification on the resulting (post-FEC)  
> shares in less time than doing FEC locally. If you assume that there  
> are more backups than restores (and restores require only 1/4 the  
> bandwidth of backups), then storage servers will have unused outgoing  
> bandwidth. A FEC server OTOH has more outgoing than incoming, so  
> perhaps intermixing the two services on the same nodes would be  
> effectively free.
> 
> --Wes
> 

------- End of Forwarded Message




More information about the tahoe-dev mailing list