A Decentralized Directory Service for Peer-to-Peer-Based Telephony - Cloud, Grid and High Performance Computing: Emerging Applications - page 342

queries on top of Distributed Hash Tables. The
design of the algorithm is driven by the require-
ments found in a distributed user directory for a
commercial VoIP communication platform devel-
oped by Siemens. We evaluated the algorithm and
showed how to choose the parameters in order to
achieve the best performance.
While this article is focused on a specific use
case, the methodology and results can be trans-
ferred to other scenarios. The algorithm presented
here fits specifically in situations where keywords
are Zipf-distributed. In the phone book scenario,
some last names are very common while other last
names are very rare. The EPHT adapts perfectly
to this kind of distribution.
The concatenation of keywords provides a
simple but powerful approach to handle multiple
keywords that are ordered in a hierarchical way.

based approach for registration and lookup carries
intrinsic limitations as regards the expressiveness
of search. While the extension with range queries
and wildcard search seems appropriate for a pure
phone book lokup, even a straightforward business
directory will require more semantically elabo-
rate queries (e.g., SQL-based or ontology-based
queries). One option to achieve this is to combine
structured distributed hash tables with super-peer
architectures, preserving the robustness and scal-
ability of the overlay while enhancing it with
declarative semantic search capability. In (Gerdes
et al., 2009), we propose a declarative decentral-
ized query processor and evaluate it in the energy
domain. (Stiefel and Müller, 2010) propose the use
of an ontology-based query language on top of a
DHT architecture for semantic search of digital
product models. These approaches will need to be
validated and further developed in future work.

FUTURE RESEARCH DIRECTIONS

REFERENCES

The Extended Prefix Hash Tree algorithm pre-
sented in this paper enables the implementation
of a distributed user directory for a peer-to-peer-
based telephony application. However, apart from
user directories, there are more applications that
might benefit from a distributed search index.
The evaluation in this article is based on
the specific requirements that we derived from
a commercial communication platform. When
EPHTs are to be applied in other applications, it
is a non-trivial task to tell the implications of the
algorithm on the specific architecture.
Future research should address this issue and
allow for the definition of generic, re-usable com-
ponents that can be applied on top of peer-to-peer
networks. These components are the building
blocks fulfilling the application-specific require-
ments on the distributed infrastructures. A first
proposal for the definitions of these components
can be found in (Stäber, 2009).
Also, while DHT-based structured overlay
networks have many advantages, their string-

Aspnes, J., & Shah, G. (2003). Skip graphs. In
SODA '03: Proceedings of the Fourteenth Annual
ACM SIAM Symposium on Discrete Algorithms ,
(pp. 384-393). Philadelphia, PA, USA.

Barsanti, L., & Sodan, A. (2007). Adaptive job
scheduling via predictive job resource allocation.
In Proceedings of Job Scheduling Strategies for
Parallel Processing (pp. 115-140).

Bellare, M., & Rogaway, P. (1993). Random
oracles are practical: A paradigm for designing
efficient protocols . In CCS '93 1st ACM Confer-
ence on Computer and Communications Security,
(pp. 62-73). New York, NY: ACM Press.

Bharambe, A. R., Agrawal, S., & Seshan, S. (2004).
Mercury: Supporting scalable multi-attribute
range queries. In SIGCOMM Symposium on
Communications Architectures and Protocols,
(pp 353-366). Portland, OR, USA.