Session 2d entitled “Future Internet Architecture” hosted presentations from academic, research and industrial institutions, covering a broad range of research topics related to Future Internet. Its purpose was to allow researchers present their ideas and get feedback from the audience on converged networks and management systems, experimentation for new protocols and socioeconomic aspects of the Future Internet.
Key themes from the session were an increasing need for:
1) treating control and data packets separately, even though the overhead caused must be carefully considered
2) handling mobility and multi-homing seamlessly, even though routing protocols must be carefully designed to handle frequent changes
3) testing the harmonic operation of multiple protocols
4) understanding the socio-economic aspects of Future Internet technologies
Similarly, Workshop 3d: Cloud Networking hosted presentations and a panel from academic and research institutions on the technical and business challenges of converged cloud computing and networking.
Bartosz Belter, from Poznan Supercomputing and Networking Center (Poland) and the GEYSERS project explained the need for having converged infrastructures (network and IT) and described their approach of using testbeds for evaluating the innovation and demonstration impact in a wide range of use cases.
Daniel Turull from KTH Royal Institute of Technology, Sweden and SAIL project described Libnetvirt, a network virtualization library for linux systems, since although virtualization technologies exist, these are very technology dependent and not suitable for dynamic environments.
Markus Fiedler from Blekinge Institute of Technology (Sweden) gave a presentation on performance of virtualized routers by examining whether isolation is feasible on different timescales using XEN hypervisor technology. He concluded that sharing of infrastructures has negative effects on performance isolation of different virtual resources and we could increase isolation but this would depend on the application requirements.
Finally, Anna Tzanakaki, AIT (Greece) in her presentation focused on how an operator should use the virtual infrastructures when dealing with network and IT resources. The optimal planning scheme was found to be the joint consideration of network & IT (instead of a simpler approach based on the location of a single infrastructure type).
The panelists’ statements revealed that the cloud technology as such serves a quite good standing, however, that the service quality levels vary depending on the layer. In addition, the level of security, the non-availability of accounting technology for virtualized environments, and the lack of control for cloud users seem to determine major drawbacks as of today.
In terms of business models the demand for user quality-related contracts as well as clear price-cost-based relation for offered services are demanded, but not available. Real guarantees for services offered need to be reflected in appropriate business models, which need to inter-relate technology-specific solutions with operational costs of clouds.
The discussion touched upon various issues, especially, the energy-saving argument for cloud services – but different panelists’ statements indicated that an optimal solution of utilization as well as energy savings may not be achievable, mainly due to the lack of realistic models for cloud services in terms of their technical needs (bandwidth, storage, usage behavior) and relevant cloud service models (which users need them at which location in the world and a which time). As such, energy-related questions can be answered only for dedicated use cases and well-defined scenarios.
Additionally, the “closer” metric for data and users in a cloud sees two different replies: (a) “closer” is driven the domain (legal, administrative), which determines the most remote location data in the cloud may see due to specific application demands and (b) “closer” is depending on the response time a user requires for a certain service.
The relevance of FuNeMS 2012 for SESERV’s was visible in a number of important topics. Firstly, the role of business models (thus, the economics of services and future networks) was clearly visible in Session 2d as well as the cloud workshop’s presentations and panel. Quoting David Kennedy, the chairman of Session 2d, “[…] all, finally, come down to the pockets of the stakeholders“ verified the community’s interest in understanding the socio-economic dimension of Future Internet architectures. Secondly, the importance of the user (determining a human being, thus, a social individual) revealed that user-perceived Quality-of-Experience (QoE), user control besides providers’ control and operational robustness, and security mechanisms for cloud-based service offerings in a fully decentralized, potentially world-wide service cloud are highly demanding. Nevertheless, the need for observing legal boundaries and state lines for certain cloud services as well as the lack of convincing Service Level Agreements with valuable, meaningful, and accountable parameters form still major obstacles for a number of commercially applicable services for companies. This seems to be less important these days for mass market cloud services for private end-user demands, such as for picture storage, social media, and smaller computing demands.