. Ntk, NDN/CCNx some critical observations
. Introduction
NDN (Named Data Networking) [1,2] is an Information Centric Network
architecture. The aim of the NDN project is "to arrive at a design of
a new network architecture for the Internet that incorporates content
delivery in its primitive functions while at the same time removing
the need of relying on costly third-parties like CDNs [Content
Distribution Networks] or on inefficient P2P systems"[3].
Both Netsukuku and NDN have ambitious goals. NDN aims to be a
next-generation Internet architecture, while Netsukuku aims to bring
forth an alternative socio-economic model for the Internet.
For what regards NDN, the NDN project has gained international
attention from academic research and private research of various
tele-communication companies[4-11].
The Netsukuku project was born and developed as a grass-root
hacktivist project.
In this document you can find thoughts and observations about how
the NDN/CCNx project could benefit the Netsukuku project.
[1] http://www.named-data.net/
[2] http://en.wikipedia.org/wiki/Named_data_networking
[3] Inter-Domain Named Data Networking, Andrea Lo Pumo, 2011
[4] Named Data Networking. http://www.named-data.net, 2012.
[5] CAIDA. Named Data Networking (NDN). http://www.caida.org/ projects/ndn-fia/, September 2012.
[6] AsiaFI. Asia Future Internet ICN working group. http://www.asiafi. net/org/ndn/, October 2011.
[7] ANR. Content oriented networking (CONNECT). http://www.anr-connect.org, 2011.
[8] PARC. PARC content-centric networking. http://www.parc.com/publications/focus-area/content-centric-networking/, 2009.
[9] PARC. CCNx. http://www.ccnx.org, 2012.
[10] A. Afanasyev, I. Moiseenko, and L. Zhang. ndnSIM: NDN simulator for NS-3. Technical report, Tech. Rep. NDN-0005, NDN Project (July 2012). URL http://irl.cs.ucla.edu/ndnSIM.html.
[11] A. Ghodsi, S. Shenker, T. Koponen, A. Singla, B. Raghavan, and J. Wilcox. Information-centric networking: seeing the forest for the trees. In Proceedings of the 10th ACM Workshop on Hot Topics in Networks, page 1. ACM, 2011.
. Observations
. At a very high-level, Netsukuku could implement NDN features by
recurring to ANDNA[12] for dealing with names and by introducing a
Content Store for enabling caching of forwarded contents. Vice-versa,
NDN could implement Netsukuku features by adopting Netsukuku routing
protocols, which would have to be opportunely adjusted to deal with
names rather than addresses.
Both Netsukuku and NDN needs a scalable routing protocol to meet
their objectives. The Netsukuku project, made various efforts
researching a suitable routing protocol, i.e. a routing protocol
that could be dynamic and scalable up to a network of 2^32 routers
(2^32 is the number of IPv4 addresses), see [13,14,15].
On the contrary, the NDN protocol is defined by assuming the
existence of a suitable routing protocol and it does not itself
provide one. The NDN project does in fact rely on the successful
outcome of other independent research efforts regarding scalable
routing protocols[16,17,20-22].
A research about scalable routing that is independent from
considerations on how the network architecture should be implemented
is Compact Routing[18,19]. Compact Routing explores different
schemes for routing in order to achieve routing table sizes that
scale sub-linearly with the number of nodes.
For what regards Compact Routing efforts [18,19], as of summer
2010 they did not provide a dynamic compact routing scheme,
i.e. "the design of an efficient dynamic scheme that is updated
incrementally as the network evolves [was] an open problem [and
perhaps as of now (summer 2013) still is]"[13].
To some the most promising class of routing architectures
seems to be hybdrid ones where multiple servers scattered in the
network, like super-nodes in P2P networks, serve to digest the
load of the routing protocol and make it scale, see as example [16].
[12] A Netsukuku Domain Name Architecture, http://netsukuku.freaknet.org, AlpT (@freaknet.org), July 3, 2008, http://netsukuku.freaknet.org/doc/main_doc/andna.pdf
[13] The address space balancing problem, Andrea Lo Pumo, 2010, http://www.cl.cam.ac.uk/~al565/output/interests.html
[14] The QSPN - routing discovery algorithm, http://netsukuku.freaknet.org/doc/main_doc/qspn.pdf
[15] Netsukuku topology, http://netsukuku.freaknet.org/doc/main_doc/topology.pdf
[16] NDN and IP routing, can it scale? Ashok Narayanan, David Oran, Cisco (2012) http://trac.tools.ietf.org/group/irtf/trac/raw-attachment/wiki/icnrg/IRTF%20-%20CCN%20And%20IP%20Routing%20-%202.pdf (mirror) http://dl.dropbox.com/u/78619439/mirrored-papers/ndn-and-ip-routing-can-it-scale-qm.pdf
[17] D. Perino and M. Varvello. A Reality Check for Content-Centric Networking, Proc. of ACM Sigcomm ICN 2011. August 2011.
[18] Papadimitriou, D. and NV, A.L.B. Compact Routing: Challenges, Perspectives, and Beyond. http://streaming.info.ucl.ac.be/data/grascomp/pdf/TFISS09-Papadimitriou.pdf
[19] A.Korman and D.Peleg, Dynamic routing schemes for general graphs, In Proc. of ICALP 2006, Part I, LNCS 4051, pp. 619-630, Springer-Verlag Berlin Heidelberg 2006.
[20] D. Krioukov, F. Papadopoulos, M. Kitsak, A. Vahdat, and M. Bogu. Hyperbolic Geometry of Complex Networks, Physical Review E, vol. 82, no. 036106, Oct 2010.
[21] Potential based routing for ICN, Suyong Eum, Kiyohide Nakauchi, Takeshi Usui, Masayuki Murata, Nozomu Nishinaga, Proceeding AINTEC ’11 Proceedings of the 7th Asian Internet Engineering Conference Pages 116-119 ACM New York, NY, USA 2011, doi 10.1145/2089016.2089035
[22] J. Rajahalme, M. Srel, K. Visala, and J. Riihijrvi. On Name-based Inter-domain Routing. Comput. Netw., December 2010.
. Both NDN and Netsukuku leave aside MAC considerations, i.e. what
physical technologies allow nodes to exchange data (neighbours to
be linked).
The Netsukuku project, which started by considering the idea of
adopting wireless links for building autonomous MANETs (Metropolitan
Area Networks), if to be developed further should consider the idea
of researching scalable routing protocols optimized for a physical
layer of choice, e.g. wireless. An example of optimizing a network
protocol for wireless is tcp/ip designed with network coding[23].
[23] Network coding meets TCP, Jay Kumar Sundararajan, Devavrat Shah, Muriel Medard, Michael Mitzenmacher, Joao Barros, http://arXiv.org/abs/0809.5022v1
. A suggested line of action is to incorporate the core of the
NDN/CCNx protocol in one netsukuku prototype, while preserving
the compatibility with the NDN/CCNx protocol (see for example
CCNxLite[24], a light-weight implementation of CCNx, compatible
with CCNx), and developing the netsukuku routing protocols (and
other protocols like, for example, Viphilama[25-26], Carciofo[27])
on this core.
This will allow to collaborate with the NDN research community
and gain traction from the inertia that NDN research and
ICN (Information Centric Networking) more in general has
gained. Moreover, the NDN protocol is defined independently
from (almost) any particular choice of routing protocol. Thus,
the current routing protocols developed for Netsukuku are not
excluded from this line of action and can be further developed if
it is seen that they are appropriate or seem to be promising to
be researched further.
[24] CCNxLite http://www.ccn-lite.net/CCNlite-Nice2012-Poster.pdf
[25] Ntk_RFC Viphilama http://netsukuku.freaknet.org/index.php?pag=documentation&file=/main_documentation/ntk_rfc/Ntk_viphilama
[26] Ntk_RFC Viphilama Static http://netsukuku.freaknet.org/index.php?pag=documentation&file=/main_documentation/ntk_rfc/Ntk_viphilama_static
[27] Ntk_RFC Carciofo http://netsukuku.freaknet.org/index.php?pag=documentation&file=/main_documentation/ntk_rfc/Ntk_carciofo
