Periodic Reporting for period 3 - ACINO (Application Centric IP/Optical Network Orchestration)
Période du rapport: 2017-02-01 au 2018-01-31
Applications and verticals that support specific business processes, with their peculiar requirements, will ultimately drive the evolution of network technologies towards the Next Generation Internet. In the transport networking area, relevant business applications could e.g. be virtual machine migration, data backup (sensitive financial or social network data), distributed content delivery (video-on-demand, TV broadcasting), etc. In the last few years, the needs of these applications evolved from simple requirements, which can be simply met with a best-effort network and a reliable end-to-end transport protocol, to stringent ones in terms of latency and jitter (financial transactions, remote robot control), bandwidth (remote video production over IP), reliability, security, etc. These needs cannot be (cost-effectively) met using current technology.
In current transport networks, traffic generated by applications is aggregated by a grooming layer (IP/MPLS) into large flows that can be cost-effectively supported by the underlying optical layer. This grooming process leads to an indirect relationship between applications and network, and an inaccurate mapping of application needs to services provided (due to the aggressive aggregation). It is, however, prohibitively expensive to provide the best possible service to each individual application at the optical layer.
The ACINO (Application Centric IP/optical Network Orchestration) project developed and demonstrated a dynamic, application-centric, multi-layer SDN network orchestrator, where each application receives a tailored service at each layer of the transport network, overcoming the requirement visibility limitations introduced by the grooming layer, thus making the network more efficient and effective. Applications can even be engineered to take advantage of this personalized treatment. This caters to the needs of businesses deploying emerging medium-large bandwidth-consuming applications, while smaller flows that cannot hope to match the granularity of an optical service are still groomed together with other flows having similar requirements. The high-level ACINO concept is outlined in Figure 1.
In current transport networks, traffic generated by applications is aggregated by a grooming layer (IP/MPLS) into large flows that can be cost-effectively supported by the underlying optical layer. This grooming process leads to an indirect relationship between applications and network, and an inaccurate mapping of application needs to services provided (due to the aggressive aggregation). It is, however, prohibitively expensive to provide the best possible service to each individual application at the optical layer.
The ACINO (Application Centric IP/optical Network Orchestration) project developed and demonstrated a dynamic, application-centric, multi-layer SDN network orchestrator, where each application receives a tailored service at each layer of the transport network, overcoming the requirement visibility limitations introduced by the grooming layer, thus making the network more efficient and effective. Applications can even be engineered to take advantage of this personalized treatment. This caters to the needs of businesses deploying emerging medium-large bandwidth-consuming applications, while smaller flows that cannot hope to match the granularity of an optical service are still groomed together with other flows having similar requirements. The high-level ACINO concept is outlined in Figure 1.
ACINO first characterized the requirements of multiple network business applications, then identified the multi-layer network operations needed to support them, the required technology and reference network. The use cases validating the project also stemmed from this study.
Then, the requirements and architecture of the multi-layer network orchestrator were defined (a simplified schema is provided n Figure 2). The project defined a grammar to enable applications to abstractly describe the service they need, implemented in a controller-agnostic intent-based interface called DISMI.
To translate such intents into configuration, ACINO developed a multi-layer application-centric algorithmic framework in Net2Plan, an open-source optimization tool. ACINO’s policies operate on many parameters (latency, availability, security, etc.) and include online (in-operation) planning functionalities.
The ACINO orchestrator was built on top of ONOS, an open-source network controller, which was modified to expose a disaggregated multi-layer network model, cooperate with Net2Plan, and support the operations defined in the project. New southbound drivers, based on existing and emerging standards (e.g. the ONF Transport-API) were implemented manage the IP and optical layers, and also introduced by ACINO industrial partners in their products. Extensive interoperability tests and demonstrations of the developed technologies were carried in collaboration with major international vendors and operators.
3 testbeds (1 emulated, 2 physical), were setup to demonstrate the project’s 5 use cases plus 3 additional demonstrations.
These showcased the qualitative benefits of the ACINO orchestrator, including the capability to provide on-the-fly secure services at different network layers (IPSec, MACSec or Optical Sec) while also respecting other service requirements (latency, throughput and payload overhead) and accounting for the network’s state, and the capability to opportunistically provide the most suitable reactive (slow) or proactive (fast) survivability option for services with different availability requirements.
ACINO’s application-aware concept was evaluated by means of techno-economic analyses. It was demonstrated to outperform application-unaware ones in terms of density of satisfiable services, admitting slightly less services but ensuring their requirements are met, but also requiring slightly more interfaces. The CAPEX impact, however, is mitigated by the opportunistic optimization of resources at both IP and optical layers. A cost study showed that to achieve the same performance using typical offline planning and optimization algorithms, operators would need to significantly overprovision their networks, leading to higher costs compared to ACINO’s.
The consortium also described and evolution path from current OSS to the ACINO solution, and some new business model enabled by allowing external player to provide programs for the configuration of the network by means of an App-Store.
The project outcomes were disseminated by means of multiple channels, with more than 80 dissemination activities (articles, presentations, tutorials, etc.). Moreover, the consortium was very active in the standardization bodies (IETF and ONF) and in contributing to open source communities (ONOS).
Then, the requirements and architecture of the multi-layer network orchestrator were defined (a simplified schema is provided n Figure 2). The project defined a grammar to enable applications to abstractly describe the service they need, implemented in a controller-agnostic intent-based interface called DISMI.
To translate such intents into configuration, ACINO developed a multi-layer application-centric algorithmic framework in Net2Plan, an open-source optimization tool. ACINO’s policies operate on many parameters (latency, availability, security, etc.) and include online (in-operation) planning functionalities.
The ACINO orchestrator was built on top of ONOS, an open-source network controller, which was modified to expose a disaggregated multi-layer network model, cooperate with Net2Plan, and support the operations defined in the project. New southbound drivers, based on existing and emerging standards (e.g. the ONF Transport-API) were implemented manage the IP and optical layers, and also introduced by ACINO industrial partners in their products. Extensive interoperability tests and demonstrations of the developed technologies were carried in collaboration with major international vendors and operators.
3 testbeds (1 emulated, 2 physical), were setup to demonstrate the project’s 5 use cases plus 3 additional demonstrations.
These showcased the qualitative benefits of the ACINO orchestrator, including the capability to provide on-the-fly secure services at different network layers (IPSec, MACSec or Optical Sec) while also respecting other service requirements (latency, throughput and payload overhead) and accounting for the network’s state, and the capability to opportunistically provide the most suitable reactive (slow) or proactive (fast) survivability option for services with different availability requirements.
ACINO’s application-aware concept was evaluated by means of techno-economic analyses. It was demonstrated to outperform application-unaware ones in terms of density of satisfiable services, admitting slightly less services but ensuring their requirements are met, but also requiring slightly more interfaces. The CAPEX impact, however, is mitigated by the opportunistic optimization of resources at both IP and optical layers. A cost study showed that to achieve the same performance using typical offline planning and optimization algorithms, operators would need to significantly overprovision their networks, leading to higher costs compared to ACINO’s.
The consortium also described and evolution path from current OSS to the ACINO solution, and some new business model enabled by allowing external player to provide programs for the configuration of the network by means of an App-Store.
The project outcomes were disseminated by means of multiple channels, with more than 80 dissemination activities (articles, presentations, tutorials, etc.). Moreover, the consortium was very active in the standardization bodies (IETF and ONF) and in contributing to open source communities (ONOS).
ACINO, through its demonstrations, proved the feasibility and advantages of the application-centric approach to multi-layer networking, while furthering the development of multi-layer SDN control solutions for modern transport networks.
With the ACINO application-centric approach, Verticals and applications owners will benefit from a transport ecosystem that can “talk their language”. This will allow service providers to strengthen their position and provide value-added services, increasing their profitability. Thanks to the better ability to control service characteristics, this will ultimately affect the experience of end users.
Simultaneously, network operators will be able to transfer, protect from failures, and secure large amounts of traffic in a more effective and automated way. They will be able to enhance their service offerings, thanks to the ability to satisfy strict requirements while avoiding service level agreement violations. This implies reduced CAPEX, due to not needing to evolve the infrastructure to satisfy such requirements, as well as OPEX, due to the reduction in manual operations needed to setup services enabled by ACINO. Furthermore, the use of standard interfaces in the project will foster a more competitive and open market, where operators can diversify risks.
Since ACINO is based on state of the art packet and circuit network capabilities (adaptive, flexible networks), the success of the project will foster the deployment of such technologies, giving European system and component vendors an advantage in creating an alternative food chain geared towards flexible and open solutions.
The knowledge and experience developed within the project on the topic of multi-layer and application-centric network control will also give control plane vendors and SDN application developers an early advantage in developing and evaluating the effectiveness and commercial viability of technical solutions within these areas.
With the ACINO application-centric approach, Verticals and applications owners will benefit from a transport ecosystem that can “talk their language”. This will allow service providers to strengthen their position and provide value-added services, increasing their profitability. Thanks to the better ability to control service characteristics, this will ultimately affect the experience of end users.
Simultaneously, network operators will be able to transfer, protect from failures, and secure large amounts of traffic in a more effective and automated way. They will be able to enhance their service offerings, thanks to the ability to satisfy strict requirements while avoiding service level agreement violations. This implies reduced CAPEX, due to not needing to evolve the infrastructure to satisfy such requirements, as well as OPEX, due to the reduction in manual operations needed to setup services enabled by ACINO. Furthermore, the use of standard interfaces in the project will foster a more competitive and open market, where operators can diversify risks.
Since ACINO is based on state of the art packet and circuit network capabilities (adaptive, flexible networks), the success of the project will foster the deployment of such technologies, giving European system and component vendors an advantage in creating an alternative food chain geared towards flexible and open solutions.
The knowledge and experience developed within the project on the topic of multi-layer and application-centric network control will also give control plane vendors and SDN application developers an early advantage in developing and evaluating the effectiveness and commercial viability of technical solutions within these areas.