Short name: LISCIo
Long name: Lightweight Self-adaptive Cloud-IoT Monitoring across Fed4FIRE+ Testbeds
Company: Department of Computer Science, University of Pisa
Call: F4Fp-08 (see call details)
Proposal number: F4Fp-08-M30
SUMMARY REMARKS & TESTBEDS
Monitoring will be crucial to properly orchestrate next-gen services. Indeed, monitoring’s output can be exploited to choose where to deploy application services for the first time and to decide when and where to migrate them in case their QoS and contextual requirements cannot be satisfied by the current deployment and infrastructure state. However, only a few works have focused so far on the design and prototyping of monitoring tools for next-gen Cloud-IoT computing platforms. In this context, FogMon (https://github.com/di-unipi-socc/FogMon) is an open-source C++ distributed monitoring service targeting heterogeneous infrastructures along the Cloud-IoT continuum, e.g. Fog computing. FogMon monitors hardware and virtualized resources at different Cloud-IoT computing nodes, end-to-end network QoS between such nodes, as well as available IoT devices. Besides, it features a self-organising peer-to-peer overlay topology with self-restructuring mechanisms and differential monitoring updates, which feature scalability, fault-tolerance, and low communication overhead. Notably, in small-scale testbed settings (13 nodes in Pisa, Italy), FogMon featured a very limited footprint on hardware and network bandwidth usage. This project aims at testing, assessing and tuningFogMon in larger-scale settings, over at least two testbeds within the Fed4FIRE+ federated infrastructure. Particularly, we aim at realising and testing a Docker-based deployment of FogMon running upon a portion of the highly distributed heterogeneous facilities provided by Fed4FIRE+ (i.e. wired, wireless and 5G connections, and Cloud and Edge nodes and IoT devices), spanning from 20 to40+computing nodes. We also plan to exploit the FedMon monitoring tool of Fed4FIRE+to assess the accuracy of the results collected by FogMon for those metrics that are collected by both services (e.g. resource usage, latency). We expect that the final release of FogMon could also be exploited by other Fed4FIRE+ experimenters to enable the QoS-and context-aware deployment of their applications on a selected infrastructure slice.