[chord] simulator query

[istoica at EECS.Berkeley.EDU]

Tim,

Sorry for the delay in answering your question, but I was
on vacation. Anyway, here is the answer.

Actually, in terms of functionality the simulator code is not that 
different from the Chord pseudocode. The confusion may occur because 
the fingerList data structure in the simulator includes three data 
structures: (1) the successor list, (2) the list of proper fingers (i.e., 
fingers at the distance 2^k), and (3) a cache of fingers. 

While the fingers in fingerList are replaced using LRU, note that the
fingers in the successor list and the proper fingers are refreshed at
a much higher rate than the other fingers (i.e., the cached fingers) so in
the vast majority of cases LRU will only replace these cached fingers.

The advantage of this implementation is that you need to consult only
one data structure (i.e., fingerList) when you answer a query. The
disadvantage is of course its readability.

Hope this helps. If you have more detailed comments/questions you 
can send them directly to me. Best,

Ion



----- Original Message -----
From: "Thomer M. Gil" <thomer at lcs.mit.edu>
Date: Sunday, July 27, 2003 5:27 pm
Subject: Re: [chord] simulator query

> ----- Forwarded message from Tim Hu <timhu at mobqos.ee.unsw.edu.au> -
> ----
> 
> From: Tim Hu <timhu at mobqos.ee.unsw.edu.au>
> To: "Thomer M. Gil" <thomer at gil.com>
> Date: Mon, 28 Jul 2003 10:25:49 +1000 (EST)
> Subject: Re: [chord] simulator query
> 
> dear thomer:
> 
> sorry for the delayed reply, was on holiday ...
> 
> > Could you tell us what it is that makes you say this?  Where do you
> > think the implementation deviates from the ``specifications''?
> 
> in the chord simulator, the finger table entries is refreshed by a LRU
> policy, and doesn't seem to conform to key-space routing (as said 
> in the
> chord paper).
> 
> by key-space routing i mean each entry in the finger table is 2^x 
> keyspaceaway from the node, and then you match the closest nodeid 
> to the finger.
> 
> > We are currently working on another simulator that we will 
> release in
> > September.
> 
> eagerly awaits,
> 
> thank you,
> 
> tim
> 
> ----- End forwarded message -----
> _______________________________________________
> chord mailing list
> chord at amsterdam.lcs.mit.edu
> https://amsterdam.lcs.mit.edu/mailman/listinfo/chord
>