218 |
|
|
219 |
\section{Centralized} |
\section{Centralized} |
220 |
|
|
221 |
Napster \cite{napsterurl} \footnote{We decided to include Napster in this section only because it has |
Napster\footnote{We decided to include Napster in this section only because it has |
222 |
historical value (see previous section).} was designed to to allow people to share music. |
historical value (see previous section).} \cite{napsterurl} was designed to to allow |
223 |
It was a hybrid Peer-to-Peer file-sharing system, i.e., the search index was centralized |
people to share music. It was a hybrid Peer-to-Peer file-sharing system, i.e., the search |
224 |
and the distribution storage and serving of files was distributed. Peers in the Napster |
index was centralized and the distribution storage and serving of files was distributed. |
225 |
network performed requests to the central directory server to find other peers hosting |
Peers in the Napster network performed requests to the central directory server to find |
226 |
desirable content. Since service requests was totally based on centralized index, |
other peers hosting desirable content. Since service requests was totally based on |
227 |
Napster didn't scale well because of constantly updated central directory, and had a |
centralized index, Napster didn't scale well because of constantly updated central |
228 |
possibility to single point of failure. |
directory, and had a possibility to single point of failure. |
229 |
|
|
230 |
|
|
231 |
\section{Loosely structured} |
\section{Loosely structured} |
244 |
forwards the query to their neighbors. This leads in the situation where number of messages |
forwards the query to their neighbors. This leads in the situation where number of messages |
245 |
in the network can grow with $O(n^{2})$, where $n$ is the number of participating peers in the |
in the network can grow with $O(n^{2})$, where $n$ is the number of participating peers in the |
246 |
Gnutella network. To limit the amount of network traffic, Gnutella uses Time-To-Live-limited |
Gnutella network. To limit the amount of network traffic, Gnutella uses Time-To-Live-limited |
247 |
(TTL) flooding to distributed queries. Therefore, only peers that are TTL hops away from the |
(TTL) flooding to distributed queries. Gnutella uses a breadt-First traversal with depth limit |
248 |
query originator will forward the query or respond to the query. |
$T$ (e.g., 7), where T is the system-wide maximum TTL of a message in hops. Therefore, only peers that |
249 |
|
are TTL hops away from the query originator will forward the query or respond to the query. |
250 |
|
In Gnutella network, search results are fast, because breadt-First traversal sends queries to |
251 |
|
every possible neighbor. On the other hand, this method wastes resources and doesn't scale well. |
252 |
|
|
253 |
|
According to \cite{lv02searchreplication}, Gnutella's way to perform data lookups, \emph{flooding}, has |
254 |
|
following limitations. First, choosing the approriate TTL in practice is not easy. If the |
255 |
|
TTL is too high, query originator may unnecessarily strain the network. If the TTL is too |
256 |
|
low, the query originator might not find the desired data even it's available somewhere |
257 |
|
in the network. Second, there are many duplicate messages generated by flooding, especially |
258 |
|
in high connectivity graphs. It is obvious that with these limitations, flooding creates |
259 |
|
significant message processing overhead for each query. Furthermore, as a result, |
260 |
|
flooding may increase the load on participating to the point, where it has to leave the network. |
261 |
|
|
|
|
|
262 |
|
|
263 |
Recently, however, there has been done research on topology properties of the Internet \cite{adamic99small} |
|
264 |
and the Gnutella network \cite{adamic02localsearch}, \cite{adamic01powerlawsearch}. Studies show |
Lately, there has been done lot of research to improve Gnutella's data lookup efficiency |
265 |
that both networks has a power law distribution of links, i.e., a few peers have high connectivity |
and scalability. Adamic et. all \cite{adamic99small}, \cite{adamic02localsearch}, |
266 |
and major of peers have low connectivity. |
\cite{adamic01powerlawsearch} has been studied different random walk methods in power-law |
267 |
peers prefential attach |
networks\footnote{In power-law networks only a few peers have high number of neighbor |
268 |
to popular peers |
links and major of peers have low nuber of neighbor links.} and they have found that by |
269 |
|
instructing peers forwarding queries to select high degree peers the data lookup's |
270 |
|
performance increases signficantly. However, it's not clear whether this algorithm |
271 |
|
is scalable or not. |
272 |
|
. |
273 |
|
|
274 |
|
|
275 |
|
|
276 |
|
|