General Context
Cloud services were originally deployed over the top, in particular over the Internet. Market analysts have shown that a broader adoption of cloud services by enterprises requires Guaranteed QoS, Resilience, Security, Portability, Reversibility and Compliance with enterprise external and internal policies, e.g. related with the location of the data.
Providing evolved cloud SLAs requires a major architecture evolution and probably new business models. As a consequence, the whole eco-system is changing. Telecom Service Providers (TSPs) are deploying their own cloud services. Several TSPs have acquired cloud service providers in recent years. In spite of the potential end-to-end control, TSPs don’t have so far an integrated control plane and lack tools for the end-to-end monitoring of the services. TSPs are also extending their VPN services to enable their customers to have direct access to both, their own cloud services and cloud services provided by 3rd parties. Nevertheless, several limitations of present architectures limit the scope of potential services.
New network service providers are arriving to the market with a very focused positioning: to bring cloud services from 3rd parties closer from their customers.
In order to provide global coverage, various players are foreseeing the possibility of federating their solutions. This raises several technical and architectural issues.
OTTs are deploying data centres in various regions in order to enhance quality by keeping the independence from other players. In some cases OTTS are deploying their own network infrastructure, for example to interconnect their data centres.
In the mid-long term, we will see cloud services spread out of the big data centres that host them today. In integrated network-cloud architectures, part of the cloud infrastructure will be deployed in TSPs point of presence and possibly even in CPEs and mobile devices (smartphones, tablets…). The concept of Mobile Cloud Computing is evolving. It started as an efficient access to cloud services from mobile devices and it is migrating towards a global architecture where those devices become an integral part of the cloud infrastructure, providing at least storage and possibly local computation resources for a distributed implementation of cloud applications.
Research challenges
The work carried out during the thesis will deal will the following challenges:
- Define relevant future use cases and corresponding requirements.
- Clean slate design of future network-cloud architecture (benchmark) as well as potential migration roadmaps. Introduce the concept of “broker” in the global architecture.
- Special focus on specific functionality design related with orchestration and end-to-end provisioning of the services, resources allocation and optimization, virtual machines mobility, monitoring, etc. Design of the required resource optimization mechanisms, covering jointly network and cloud resources. Design of other optimization mechanisms, such as content location over distributed clouds (as opposed to data centre-based clouds).
- Special focus on interoperability, federation and reversibility
- Evaluating the benefits of network virtualization as an enabler for future services definition, better network/cloud integration and flexible system operation.
- Proposal and performance analysis of pricing models to clients and of evolved business models among the players.
Candidates are invited to send their CV to Prof. Daniel Kofman at daniel.kofman@telecom-paristech.fr
Proposed by:
- Jerome Dilouya, Founder and CEO of InterCloud
- Daniel Kofman, Director of the LINCS