Towards Taxonomy-based
Routing in P2P Networks
Alexander L¨oser
指導老師:許子衝 老師
學生:羅英辰
學號:M97G0216
1
Introduction(1)

The development of smart, scalable approaches
for the discovery and location of data sources in
distributed heterogeneous information systems
is an important problem in many scientific and
commercial domains.

In the e-learning domain during the last years a
large number of digital e-learning repositories
has been build.
2
Introduction(2)
3
Super-Peer based Architecture(1)

A super-peer is a node that acts as a centralized
server to a subset of clients,e.g. information
provider and information consumer.

Super-peers are also connected to each other
as peers in a pure system are (Figure 2), routing
messages over this overlay network, and
submitting and answering queries on behalf of
their clients and themselves.
4
Super-Peer based Architecture(2)
5
Models and Queries(1)

Each peer is classified by paths in one or more
taxonomies and publishes a model based on
semi-structured XML data with the taxonomies
and paths.

Open Directory Project (ODP) 是網站的開放內
容目錄，也就是所謂的 DMOZ (來自其原始網域名
稱：directory.mozilla.org)。
6
Models and Queries(2)
7
Models and Queries(3)

To lookup peer models we use a subset of the
XPATH(XML Path) language.
8
Distributed Hash Tables (DHT)
9
Indexing Peer Models and Taxonomies in a DHT(1)



Models are indexed in a catalog based on a
Distributed Hash Table.
The Catalog is distributed among the SP-SP
network.
Consider the model with PID=E.
10
Indexing Peer Models and Taxonomies in a DHT(2)
 Use
SHA-1(Secure Hash Algorithm)
11
Indexing Peer Models and Taxonomies in a DHT(3)

SUCC (successor)
12
Indexing Peer Models and Taxonomies in a DHT(4)
13
CHORD protocol
14
Indexing Peer Models and Taxonomies in a DHT(5)

Keys are stored clockwise at the closest node with the
next higher hash value.
15
Lookup Models in a DHT(1)
Exact Lookups
 BFS-based Lookups
 Conjunctive Lookup

16
Lookup Models in a DHT(2)

Exact Lookups

Figure 4.
q1/Computers/Programming/Languages/Java.

The taxonomy path of the query is hashed to $EA66
and then a lookup on the Chord ring is executed.

The result of the lookup is a set of PIDs storing
models with this classification path, e.g the peers with
the PID: D,E,F.
17
Lookup Models in a DHT(3)

BFS-based Lookups
18
Lookup Models in a DHT(4)

Conjunctive Lookup

Ex: Figure 4 q3
19
Storage Load Balancing Strategies
20
Implementation and Evaluation(1)
50 Super-peer
15000 Peers
Join and leave
within 3600s




Without load balancing(-VS-LBM)
Virtual server(+VS)
Partition based load balancing(+LBM)
Combination of partition based load balancing and virtual server (+LBM+VS)
21
Implementation and Evaluation(2)

Our load balancing approach performs better than virtual
server(+VS) and the simulation without any load
balancing(-VS-LBM).

This result are valid for a small super-peer network, such
as simulated in our experiment.

In our approach we are only able to reduce the number
of taxonomy paths a super-peer is responsible for.
22
Implementation and Evaluation(3)


Each peer issues each 240 sec an exact query for a
taxonomy path.
The average required bandwidth for serving queries and
joining and leaving peers each super-peer is
25KByte/sec.
23
Implementation and Evaluation(4)

Figure 10 shows the costs using our storage
load balancing approach only for joining leaving
peer nodes (J/L) and for issuing queries and
joining and leaving peer nodes (J/L +Query).
24
Implementation and Evaluation(5)

J/LJoin and Leave
25
Summary and FurtherWork

We presented a completely new approach for
enabling efficient semantic query routing in P2P
networks.

Much work remains, for example dynamic
storage load balancing strategies allowing
super-peers to join and leave the catalog with a
high frequency while the catalog remains robust.
26