Run Your Experiment on the SoftFIRE Federated Testbed

 


As shown in the picture below, the experimenter will access to the SoftFIRE infrastructure only using the jFed tool. jFed will then communicate with the FITEagle instance running in the SoftFIRE testbed. FITEagle is preconfigured in order to be able to interface with the northbound API of an Open Baton instance. Finally Open Baton will be in charge of deploying the requested NFV experiments in the SoftFIRE platform.

SFArchitecture

The experiment lifecycle is thus divided in different phases.

  • Access to the federated testbed
  • Design and Programming
  • Discovery and Allocation of resources
  • Running and troubleshooting the experiment

Access to the federated testbed

Once the experimenter gains the right to execute the experiments in the SoftFIRE portal, his first step is to retrieve his private certificate that will allow him to execute the requests for gathering the available resources and for uploading his experiment software. It will be possible to retrieve it from this site after the open call will be closed. This certificate will also be used in the next phase in order to access the SoftFIRE VPN and be able to connect to his specific resources.

Design and Programming

Gained access to the platform, the experimenter can start to develop his experiment (if this process was not yet started). This include the creation of a particular cloud image containing the software to be executed during the test. We strongly suggest to test this image multiple times before executing the experiment in the SoftFIRE platform in order to reduce the troubleshooting time. The image upload could take several minutes. When the image is ready, the experimenter needs to create some packages containing the scripts that will install, configure and start the software in the image. Some VNF Packages will be created and the procedure is shown in this tutorial.
Now that the image and the packages are created they can be uploaded in the SoftFIRE software portal. The image has to be uploaded first and after that an id will be returned. This id has to be set in the Virtual Network Function Descriptor that is contained in the VNF Package. In this was the VNF described by that VNF Package will use the image that has been just uploaded.  After that, the VNF Package can be uploaded.

Discovery and Allocation of resources

Once the image and the packages are uploaded, jFed will be able to do a discovery of the new resources (Virtual Network Functions) and the experimenter will be able to combine them in order to create a Network Service. After that, it will be possible to run the experiment from the jFed GUI.

Running and troubleshooting the experiment

Some minutes are needed to deploy and configure the experiment, depending on the software and on the configuration scripts. After that, if everything goes well the experiment will reach the ACTIVE status and from the jFed tool it will be possible to see the IPs assigned to the machines.
It will be possible at that time, to access the Virtual Machine passing through the OpenVPN server. In order to access the OpenVPN it is needed to use the same certificate that has been provided in the first phase. From here, the experimenter has access to all and to only his Virtual Network Functions and he or she can start the experiment (if not started from the VNF Package scripts) and gather results.
In case of occurring errors, the experimenter can access the Virtual Machine using SSH through the OpenVPN connection, check the log files of the instantiation and finding the issue. SSH login is possible using the ssh-key that was passed to jFed during the creation of the Experiment.