The adoption of virtualization/containerization technologies in industrial control systems promises improved deployment flexibility, scalability, and resource efficiency. However, migrating time-critical control logic to shared virtualized environments poses significant challenges in meeting strict real-time requirements. We address this challenge by evaluating the network and compute performance of containerized control deployed on commodity-of-the-shelf hardware using a real-time Linux kernel (PREEMPT_RT). The paper reports systematic measurements using resource partitioning and kernel tuning techniques to mitigate interference and improve task predictability. Through experiments measuring computing time stability and network round-trip time, we quantify the impact of virtualization and resource contention. Results show that while PREEMPT_RT introduces minor network latency overhead (e.g., a median increase of approximately 5 µs in round-trip time, with the overall Round Trip Time (RTT) remaining below 50 µs), specific kernel configurations significantly improve computing stability and, in some cases, reduce runtime standard deviation to as low as 4 µs. Crucially, we reveal that interactions between schedulers for concurrent computing and high-priority network tasks can severely degrade network predictability under high system load, by increasing jitter from near-zero to millisecond in some scenarios. Based on these findings, we provide experimentally validated guidelines for configuring commodity platforms to preserve deterministic performance in containerized industrial control.

Hofer, F., Garbugli, A., Venanzi, R., Bujari, A., Bellavista, P. (2026). Beyond bare metal: Next-generation industrial control via resource slicing on commodity platforms. JOURNAL OF SYSTEMS ARCHITECTURE, 181, 104010-104010 [10.1016/j.sysarc.2026.104010].

Beyond bare metal: Next-generation industrial control via resource slicing on commodity platforms

Bujari A.
Co-primo
Membro del Collaboration Group
;
Bellavista P.
Co-primo
Membro del Collaboration Group
2026

Abstract

The adoption of virtualization/containerization technologies in industrial control systems promises improved deployment flexibility, scalability, and resource efficiency. However, migrating time-critical control logic to shared virtualized environments poses significant challenges in meeting strict real-time requirements. We address this challenge by evaluating the network and compute performance of containerized control deployed on commodity-of-the-shelf hardware using a real-time Linux kernel (PREEMPT_RT). The paper reports systematic measurements using resource partitioning and kernel tuning techniques to mitigate interference and improve task predictability. Through experiments measuring computing time stability and network round-trip time, we quantify the impact of virtualization and resource contention. Results show that while PREEMPT_RT introduces minor network latency overhead (e.g., a median increase of approximately 5 µs in round-trip time, with the overall Round Trip Time (RTT) remaining below 50 µs), specific kernel configurations significantly improve computing stability and, in some cases, reduce runtime standard deviation to as low as 4 µs. Crucially, we reveal that interactions between schedulers for concurrent computing and high-priority network tasks can severely degrade network predictability under high system load, by increasing jitter from near-zero to millisecond in some scenarios. Based on these findings, we provide experimentally validated guidelines for configuring commodity platforms to preserve deterministic performance in containerized industrial control.
2026
Hofer, F., Garbugli, A., Venanzi, R., Bujari, A., Bellavista, P. (2026). Beyond bare metal: Next-generation industrial control via resource slicing on commodity platforms. JOURNAL OF SYSTEMS ARCHITECTURE, 181, 104010-104010 [10.1016/j.sysarc.2026.104010].
Hofer, F.; Garbugli, A.; Venanzi, R.; Bujari, A.; Bellavista, P.
File in questo prodotto:
Eventuali allegati, non sono esposti

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11585/1092730
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? 0
  • OpenAlex ND
social impact