Abstract: Large-scale production systems are largely operated by management software like Kubernetes operators and increasingly by AI agents. Production systems have diverse functionalities, interfaces, and fault-tolerance properties; hence, their management semantics are complex and less standard. Today, management-plane software is largely implemented based on experience and documents of the managed systems, where correct management is a best effort. We advocate for the practice of writing formal operations specification (FOS) as a formal contract to precisely describe expected behaviors of the production system and to enable systematic checking and verification of management-plane software and agents. We argue that the FOS should be part of standard system interface to enable reliable system management. We suggest that every production system include a FOS and share our experience of writing FOSes for real-world systems; we show that writing and maintaining a FOS is affordable and can potentially be automated
DevOps
Formal Operations Specifications for Reliable System Management
Lobsters0 views
Source:https://djapg7v42ea9q.cloudfront.net/pdf/hotnets26-final281.pdf
Related posts

I Deployed a Game to AWS by Uploading One .zip File (S3 + Lambda + SSM + EC2)
0dev.to AWS

aerynOS Gets Linux 7.3.4, NVIDIA 620 Drivers, and systemd 262
0Linuxiac
Distribution Releases: AlmaLinux OS 10.3, 9.9
0DistroWatch
MLOps Explained for Software Developers: CI/CD for AI in 2026
0dev.to Java

A governed AI-driven development lifecycle for financial services with Kiro and AWS Frontier Agents
0AWS DevOps Blog
Test a Linux health warning without filling your disk
0dev.to DevOps