Authors : Yi Wang, Huichen Dai, Junchen Jiang, Keqiang He, Wei Meng, Bin Liu
Publisher GLOBECOM 2011
Presenter : Chai-Yi Chu
Date : 2012/12/19
1



Introduction
Parallel Name Lookup(PNL)
Evaluation
2




A name lookup architecture using parallel memory
resources.
Based on IP prefix tree, called Name Prefix Tree (NPT)
where each node represents a set of components.
Group up the states of different levels in the NPT
Store the groups to individual physical modules module.
3

NDN Names
4

NPT = Name Prefix Tree
5
6
7
8

The Parallel Name Lookup algorithm consists of three
sub-procedures:
◦ 1) Build NPT based on the names stored in FIB;
◦ 2) Calculate the access probability of every node in an NPT
according to the prior knowledge of each domains access
probability. If unknown, set the access probability of all the
names to the same average value;
◦ 3) Allocate all the states, including originals and duplications,
to k physical modules.
9
10
11

Domain information
◦ DMOZ
 ODP - Open Directory Project
 1450946 domains
◦ ALEXA
 Web Information Company
 1000000 domains
12
13

Simulation 1
◦ Use DOMZ
◦ Without prior access probability of names

Simulation 2
◦ Use ALEXA
◦ With prior access probability of names
14
15

Simulation results
16


When 𝑃𝑡 increases, the speedup may vary within a
range but with no explicit discipline.
When 𝑃𝑡 exceeds a threshold, the speedups of DMOZ
and ALEXA will not change any more
17