16-06-2012, 11:23 AM
On the Myths of Optical Burst Switching
[attachment=24658]
Abstract
This paper discusses merits and drawbacks of the
Optical Burst Switching (OBS) paradigm, which received so
much attention by the academic research community. Since the
topic is hot, and given the several flavors of OBS available in
the literature, we first define what we mean by OBS.
INTRODUCTION
SINCE its origins more than a decade ago [1], [2], the
Optical Burst Switching (OBS) paradigm has attracted an
enormous research interest. Most network evolution roadmaps
refer to OBS as a viable technology in the long way towards
the always far all-optical networks dream. Successive
refinements and studies have been conducted in numerous
fields that grew up related to the OBS technology.
THE OBS TESTING SCENARIO
This section is devoted to describing and justifying the OBS
testing scenario. Every effort has been put in defining precisely
the relevant network elements and engineering decisions, to
allow results to be repeatable.
The OBS network is composed of OBS interconnection
nodes and OBS edge nodes. Interconnection nodes are connected
in a mesh topology by WDM interconnection links that
can span hundreds of km. In general, each interconnection
node is connected to one edge node1, located at the same
facility, by means of short WDM links (edge links). Edge
nodes are specific equipment, in charge of assembling the
bursts from the electronic traffic demand, and injecting them
into the network. All the fibers in the links have the same
number of wavelengths.
THE OCS TESTING SCENARIO
In OCS networks, the traffic demand is carried on transparent
permanent lightpaths (or with a sufficient holding
time for being considered as permanent) which are optically
switched in the intermediate nodes. A lightpath occupies one
E/O transmitter in its initiating node, one wavelength in each
traversed fiber, and one O/E receiver in the ending node.
The traffic demand is routed on top of the lightpaths (which
make up the so-called virtual topology), and the lightpaths
are routed on top of the fiber links. For this reason, OCS
networks are commonly named as optical multilayer networks.
Traffic from a source edge node can in general traverse more
than one lightpath, being (electronically) switched from one
to another lightpath at intermediate OCS nodes.
CONCLUSIONS
This paper critically revised some of the merits of OBS as
a competing candidate to replace OCS backbone networks. In
particular, we motivate our disbelief towards two statements
commonly associated with the OBS paradigm: its ability to
deliver the optical bandwidth more efficiently than the OCS
approach, thanks to its sub-wavelength switching granularity,
and its ability to solve contention without optical buffers.
[attachment=24658]
Abstract
This paper discusses merits and drawbacks of the
Optical Burst Switching (OBS) paradigm, which received so
much attention by the academic research community. Since the
topic is hot, and given the several flavors of OBS available in
the literature, we first define what we mean by OBS.
INTRODUCTION
SINCE its origins more than a decade ago [1], [2], the
Optical Burst Switching (OBS) paradigm has attracted an
enormous research interest. Most network evolution roadmaps
refer to OBS as a viable technology in the long way towards
the always far all-optical networks dream. Successive
refinements and studies have been conducted in numerous
fields that grew up related to the OBS technology.
THE OBS TESTING SCENARIO
This section is devoted to describing and justifying the OBS
testing scenario. Every effort has been put in defining precisely
the relevant network elements and engineering decisions, to
allow results to be repeatable.
The OBS network is composed of OBS interconnection
nodes and OBS edge nodes. Interconnection nodes are connected
in a mesh topology by WDM interconnection links that
can span hundreds of km. In general, each interconnection
node is connected to one edge node1, located at the same
facility, by means of short WDM links (edge links). Edge
nodes are specific equipment, in charge of assembling the
bursts from the electronic traffic demand, and injecting them
into the network. All the fibers in the links have the same
number of wavelengths.
THE OCS TESTING SCENARIO
In OCS networks, the traffic demand is carried on transparent
permanent lightpaths (or with a sufficient holding
time for being considered as permanent) which are optically
switched in the intermediate nodes. A lightpath occupies one
E/O transmitter in its initiating node, one wavelength in each
traversed fiber, and one O/E receiver in the ending node.
The traffic demand is routed on top of the lightpaths (which
make up the so-called virtual topology), and the lightpaths
are routed on top of the fiber links. For this reason, OCS
networks are commonly named as optical multilayer networks.
Traffic from a source edge node can in general traverse more
than one lightpath, being (electronically) switched from one
to another lightpath at intermediate OCS nodes.
CONCLUSIONS
This paper critically revised some of the merits of OBS as
a competing candidate to replace OCS backbone networks. In
particular, we motivate our disbelief towards two statements
commonly associated with the OBS paradigm: its ability to
deliver the optical bandwidth more efficiently than the OCS
approach, thanks to its sub-wavelength switching granularity,
and its ability to solve contention without optical buffers.