[chord] Etna

[Emil Sit]

On Fri, 09 May 2008 at 16:42 (-0400), Stu Hood wrote:
> After reading about Etna ( dspace.mit.edu/handle/1721.1/30555 ), I
> found this thread from January inquiring about it:
> https://pdos.csail.mit.edu/pipermail/chord/2008-January/000820.html
> 
> I've heard that Paxos can be quite challenging when it leaves paper,
> but I was wondering: what was the actual reason for abandoning the
> implementation? Lack of time? Excess complexity? Problems with the
> algorithm description? 

I'm not sure about Etna; I don't think it was Paxos that led to
it being dropped but I've added the developer to the cc for the
final word.

Paxos is pretty tough to get right from everything I've heard.
You may find this recent lab from our distributed systems class
interesting: http://pdos.csail.mit.edu/6.824/labs/lab-8.html

Also, Google put out a "paper" (a bit short on truly useful
details) that you may find helpful in implementation:

    Paxos Made Live - An Engineering Perspective (2006 Invited Talk),
    Tushar Deepak Chandra, Robert Griesemer, Joshua Redstone,
    Proceedings of the 26th Annual ACM Symposium on Principles of
    Distributed Computing, 2007.  http://www.chandrakin.com/paper2.pdf


-- 
Emil Sit / MIT CSAIL PDOS / http://pdos.csail.mit.edu/chord/