Over the past year and a half, SURFnet and TNO have worked together on a software-defined networking (SDN) project. The objective was to gain greater insight into this technology. An SDN prototype was developed for this on SURFnet’s OpenFlow testbed. We used existing concepts, protocols and software for this as much as possible. The prototype we developed has given us a good insight into the status of SDN and into what is currently available. We will continue to closely monitor developments this year and will continue actively testing SDN.
CoCo, the GN3plus Open Call Project
CoCo (Community Connect) was one of the Open Call projects within the European GN3plus project. The project ran from October 2013 through to March 2015. Within Coco, an SDN prototype was developed for setting up private networks on demand. We wanted to develop a prototype that could potentially also be a service on our production network, in order to get a clear idea of the requirements involved for SDN and the network environment. The CoCo prototype that was developed makes it easy for end-users to set up private networks through a web portal, with connectivity only provided between the institutions that are part of the private network. The private networks are isolated from the Internet which gives them protection from online hacking.
Networks play an increasingly important role in scientific research
Scientific research is increasingly a collective pursuit, and focuses on data processing. This means that network services and facilities such as cloud computing and storage are also becoming increasingly important. Discussions with researchers have revealed that privacy and security concerns mean services and facilities often cannot or may not be made publicly available on the Internet. CoCo enables scientists from different research fields to share data safely without manual intervention from network administrators. In effect it provides a shared but discrete infrastructure, outside of the Internet. The figure below shows an example of such a discrete private network.
Workshop as input for CoCo
In 2014 a workshop was organised for researchers from various academic institutes, including the Wageningen University & Research Centre and the Leiden University Medical Center. They were asked what the potential applications would be for an on-demand VPN service in their scientific research. Participants highlighted the limitations of current network services, and also offered their suggestions for and expectations of the CoCo prototype. Different ongoing projects such as “Large-Scale Population Imaging Studies” and “Genome of the Netherlands” were presented, with a specific focus on what they would look for and require of network infrastructure and connectivity.
One major potential improvement was found to be the ability for end users to create on-demand connections between data sources, and storage and computing facilities. Another important requirement was an effective way of separating Internet-connected equipment and instruments. Existing VPN technology can be used for that separation, but it requires a manual set-up by experts, which then needs to be done in every network participating in the research. CoCo enables researchers to set up the required connectivity themselves. An important aspect of CoCo is therefore its user-friendliness for researchers. Better replicability of the solution for setting up connectivity between different networks in a project, as well as for new, similar projects, was seen as a major benefit. The outcomes of the 2014 workshop were used to form the basis of the design and implementation of the first CoCo prototype.
First CoCo prototype
The first CoCo prototype shows that a user with minimal understanding of communication networks can easily set up, modify and de-activate a VPN. The VPN is set up through a user-friendly web portal. The underlying network infrastructure is hidden with only an umbrella topology displayed. Users can clearly see which sites are part of the VPN.
Plans for the future
The CoCo project concluded on 31 March 2015. The project provided us with a thorough insight into the practical aspects of software-defined networking and clarified the subsequent objectives and challenges we might still achieve. SURFnet and TNO are therefore discussing a follow-up to this project in 2015.
This blog was written by Ronald van der Pol and Marijke Kaat.