LOCARN is a new concept of Transport Network that completely revisit data, control and management planes habits with the colors of "Autonomic Networking" functionalities. It relies on very simple hardware without expensive development and optimizes dynamically the resources. It offers the main features of the autonomic networking paradigm self-discovery, -healing, -optimization, -management, protection,…
LOCARN
L ow O pex and C apex A rchitecture for R esilient N etworks
A new approach for Transmission Networks
in line with "Autonomic Networking" functionalities
LOCARN, what is it?
Locarn is above all a delightful little village in Brittany
- http://lokarn.free.fr/
But LOCARN is also
- Anew concept of Transport Network that completely revisit data,
control and management planes habits with the colors of "Autonomic Networking" functionalities
- Awill to share this concept with the whole community
- Awillto build a test bed to prove the relevance of the concepts
2
LOCARN claims to
Dramatically decrease the Capex of the network
- Relies on very simple hardware without expensive development
- Optimizes dynamically the resources
Drastically decrease the Opex
- Plug&Play, no management at the transport level required
- Low protocol stack and less processing should generate energy saving
- Power consumption optimisation
Offer the main features of the autonomic networking paradigm self-*
- -discovery, -healing, -optimization, -management, protection*,…
Offer a "Carrier Grade" set of performance (scalability, protection,…)
Be agnostic (at least in the principle)of a lower Transport layer as well as
the Service Layer
Be "Post IP" oriented
- Fully agnostic of frame format used by the services
- No IP Network needed for management
3
General principles of the Data Plane
Data Plane uses Auto-forwarding features
- Very close (but simpler) to the Dynamic Source Routing used in RFC for ad- hoc networks
Each frame has in its header all the information required to forward it from
source to destination
- asequence of "ports*" meaning the output at each node
- An hop count
- Additional information such asa service identifier, QoS parameters,…
Forwarding process in 2 steps
- First read and decrement the hop count
- Then push the frame at the port* indicated in the table for the hop count (if not null e.g. port P)
- If hop count is null (e.g. port PE), address the frame regarding its
service identifier
* Regarding the used technologies,ports may have various meanings
4
A very simple Data Plane
No table in transfer nodes (P-nodes)
No explicit Knowledge about the topology Ability to change the route frame by frame
It is so simple that we can envisage to
- Switch the payload at the photonic level
- Process the frame header also at the photonic layer
Simplicity should bring CAPEX saving
Expected photonic process should provide Energy savings
5
Control Plane principles
Control plane mainly operates on the...