Investigating the Impact of Various Underlying Hardware Factors on Virtualised Network Deployments

The flexibility inherent to the 6G space through its embrace of Service Based Architecture (SBA) networks means that the ability to understand and rely on the performance of a component based on a datasheet alone has become more difficult. It is becoming more and more clear that if one wants to deploy the same virtualized system more than once on potentially different underlying hardware with the expectation of identical performance, there is more that needs to be done than just ensuring virtualization parameters are consistent. This paper covers an initial exploration of certain underlying hardware factors and if, how and potentially why they affect certain types of performance in virtualized networks. The methodology involves the use of the platforms built by the 6G-SANDBOX project, where a set of predefined and consistent test cases are run while various underlying factors are changed and modified. The resulting KPIs are then compared to previously understood baselines to try and understand how the alteration of these factors affects performance. The factors in question within the scope of this paper are the presence of other virtualized deployments on the same host and the allocation of CPU cores and Memory/RAM to the various virtual machines that comprise the virtualized deployment. The result described shows that while certain factors and combinations thereof can have a marked effect on the KPIs in question, the real conclusion is the understanding of the humbling level of complexity and breadth inherent to this task. In this regard, the true results of the effort described in this paper may be the stepping stones generated for the application of future research and testing.