Ginkgo Technology & Platform

Th. Bullot — 2008-10-10T07:46:34Z

Autonomic Networking and Ginkgo Agents

Ginkgo's technology provides support of Autonomic Networking applications by means of Intelligent Agents, distributed in Network Equipments (NE) across the network.

Ginkgo's Intelligent Agents play a dual role in autonomic networking applications: (i) feeding in real-time the Knowledge Plane with the information required by the application and (ii) exploiting the distributed knowledge in the Knowledge Plane to manage in real-time (to "pilot" in Ginkgo's terms) network control mechanisms.

Ginkgo Agents and the Knowledge Plane

In Ginkgo's model, the Knowledge Plane is distributed among Ginkgo agents as a set of "Situated Views". Each Situated View represents the knowledge of an individual agent regarding the situation of the network in its neighborhood and is expected to be more important for the agent than the situation in remote locations. Furthermore, updating knowledge in individual Situated Views can more easily be done in real-time with limited amount of control traffic.

Ginkgo agents working together therefore maintain an always- up-to-date collective distributed knowledge of the overall network situation as illustrated in the figure below.

Ginkgo Agents Architecture and components

Ginkgo Agents are architected to support (i) the Knowledge Plane / Situated Views, (ii) the Piloting of network control mechanisms and (iii) administrative (?) tasks associated with deployment and maintenance of Agents.

Ginkgo Agents are made of three main types of building blocks:

The Situated View Knowledge Base Dedicated to storing the structured knowledge of the agent's Situated View.

The Behaviors, In charge of autonomic processing and working as organic components permanently adapting themselves to changes in their environment. Each behavior is a sense->decide->act loop in charge of a control function.

The Dynamic Planner In charge of orchestrating the overall functioning of Behaviors. The Dynamic Planner follows a Policy provided by the user indicating how the behaviors should take into account the dynamic changes occurring in the environment.

Behavior and Dynamic planner both collect, produce and manipulate structured knowledge stored in the Knowledge Base. This knowledge can be shared automatically across the network between several agents. The Knowledge Base follows an ontology model helping to conceive structured applications.

The Ginkgo Agent Software Platform

Ginkgo Agents are implemented as software Agents embedded into the Network Equipments. They are based onto the Ginkgo Agent Platform which is conceived to ease the development of distributed autonomic networking applications.

The Ginkgo Agent Software platform is designed to run in lightweight environments, such as small embedded systems, as well as on large equipment or PCs. It is tailored to have a small footprint while providing the necessary support for Autonomic Networking.

GMPLS Router Booster

Th. Bullot — 2008-10-10T07:46:15Z

The Ginkgo MPLS/GMPLS Switch Router Booster leverages the increasingly popular concept of Knowledge Plane (KP) and the Ginkgo Distributed Intelligent Agents technology to (i) maintain an always-up-to-date distributed knowledge of the network situation and (ii) adjust in real-time MPLS/GMPLS routing parameters across the network, to significantly improve usage of resources and Quality of (...) - Products & Services

Home Network QoS Booster

Th. Bullot — 2008-09-30T12:33:57Z

The Ginkgo Broadband Home Network QoS Booster leverages the increasingly popular concept of Knowledge Plane (KP) and the Ginkgo Distributed Intelligent Agents technology to (i) maintain an always-up-to-date distributed knowledge of the QoS on all links available on the Home Network and (ii) adjust in real-time routes across the network, to significantly improve end-user experience. - Products & Services

State Of The Art - Knowledge Plane

Th. Bullot — 2008-09-30T12:33:53Z

- Technology

State Of The Art - Autonomic Networking

Th. Bullot — 2008-09-30T12:33:51Z

- Technology

Ginkgo Technology

Th. Bullot — 2008-09-30T12:33:48Z