We believe that containers and network of containers will play a pivotal role to support scientific research in the coming years. We do see a clear use case in customized data processing, where data processing and data merging happen at different locations on behalf of a single users or application. Likewise, we see containers as handy constructs to ensure network performance and security.

Our research focus is on the creation of best practices to create such containers networks. To achieve this goal our research activity continues and expands on the work done in 2016. Our 2016 demonstration showed how containers can be planned and scheduled on multiple sites via DockerSwarm and how can a simple overlay network between them can be built using VXLAN.

The 2017 focus will be devoted to the evaluations of novel container networking technologies in regard to:

• The ease of use and network performance;
• The support for multi-tenancy and the isolation between flows;
• The overlays topologies that better match different usecases (data processing or security support).

We specifically look at a number of promising emerging technologies:

• Identifier Locator Addressing (ILA) [1]
• Cilium [2]. Cilium is expected to easily support IPv6, while at the same time offering the possibility to do policy enforcement through the use of kernel technologies;
• EVPNs[3]. EVPNs use MP-BGP for exchanging address information, avoiding the use of key-value stores.

In 2017 we will use again the OpenLab and the SURFnet OpenFlow and OpenStack testbed as fabric for our investigation. We will deliver an article (to be submitted to a relevant conference and/or journal) in Q2 and we will work again at the demonstration of our research result at the SuperComputing conference in Denver in Q4 2017.

[1] http://vger.kernel.org/netconf2015Herbert-ILA.pdf
[2] https://github.com/cilium/cilium
[3] http://events.linuxfoundation.org/sites/events/files/slides/ContainerCon2015_Akiyama_16x9_0.pdf