[Gzz-commits] journals/plans SA_2003.rst

[Tuomas J. Lukka]

CVSROOT:        /cvsroot/fenfire
Module name:    journals
Changes by:     Tuomas J. Lukka <address@hidden>        03/05/14 09:25:20

Modified files:
        plans          : SA_2003.rst 

Log message:
        P2P'

CVSWeb URLs:
http://savannah.gnu.org/cgi-bin/viewcvs/fenfire/journals/plans/SA_2003.rst.diff?tr1=1.145&tr2=1.146&r1=text&r2=text

Patches:
Index: journals/plans/SA_2003.rst
diff -u journals/plans/SA_2003.rst:1.145 journals/plans/SA_2003.rst:1.146
--- journals/plans/SA_2003.rst:1.145    Wed May 14 09:23:26 2003
+++ journals/plans/SA_2003.rst  Wed May 14 09:25:20 2003
@@ -589,29 +589,20 @@
 Efficient information sharing in peer-to-peer environment 
 '''''''''''''''''''''''''''''''''''''''''''''''''''''''''
 
-HH     
-
 Efficient resource discovery is an essential problem in Peer-to-Peer networks 
because there is 
 no centralized index where to look for information about resources. There are 
two well known 
 approaches in which all modern Peer-to-Peer systems fall: the loosely 
structured approach 
-and the tightly structured approach [hyytiala03stormp2p]_. In the loosely 
structured approach the 
-construction and the maintenance of the network is controlled loosely. In the 
loosely structured approach, 
-t he performance of resource discovery is $O(n)$ and the number of network 
packets can grow with 
-$O(n^{2})$, where $n$ is the number of participating nodes in the network. 
Tightly structured approach
-is constructed deterministically, which all participating peers have to 
follow; the topology of the
-overlay and the placement of services is controlled tightly. This feafure 
provides $O(\log{n})$
-resource discovery efficiency and the number of packets grow with 
$O(\log{n})$. However,
-the tightly structured approach may not be reliable in highly adverse 
conditions and the 
-performance may suffer (e.g., sudden network partition or hostile attack). 
-Thus, either of the approches do not support efficient resource discovery in 
-peer-to-peer environment in *adverse* conditions. 
+and the tightly structured approach [hyytiala03stormp2p]_. 
+The scaling properties of these different types of algorithms are wildly 
different:
+while the  tightly structured approach allows logarithmic 
[maymounkov02kademlia]_
+or even constant-time [gupta03kelips]_ scaling for resource discovery, the 
loosely
+structured approach scales much better w.r.t. changes in the structure and
+adversarial peers.
 
-In our research, we study mathematical models which can be used for analyzing 
the impacts of adverse 
-events in Peer-to-Peer environment. Currently such impacts are not well known 
and mathematical
-framework could help researchers to understand these events and effects 
better. Also, we 
-study algorithms which provide constant or locarithmic efficiency in 
Peer-to-Peer environment
-(e.g., [maymounkov02kademlia, gupta03kelips]_. In this case, our focus is to 
investigate 
-how well these algorithms perform in highly adverse conditions.
+Our goals in this area are to develop a mathematical framework for calculating
+the scaling properties of different algorithms in different network
+topologies and under different attacks, and to attempt to find 
+algorithms with tradeoff constants between the two extremes.