Advancements and Operations for LHC Run-3 and beyond

This paper presents recent advancements in the EOS project– the large-scale data storage system developed at CERN–during the preparation and execution of LHC Run-3. Key developments include further simplification of the service architecture, enhancements to metadata performance, and the introduction of new storage device inventory and cost/value interfaces. Additionally, we introduce a new scheduler implementation, a generated REST API derived from the gRPC protocol, and improved integration of features such as SciTags and SciTokens. We report on operational experiences, including the large-scale migration from CentOS 7 to AlmaLinux 9 across over 1000 storage nodes, and significant improvements in EOS file system check (fsck) for erasure coded files. Looking ahead to Run-4 and the High-Luminosity LHC (HL-LHC) era, we outline the planned evolution of EOS, including ongoing software R&D and storage technology evaluations. We discuss performance testing of hard disks using shingled magnetic recording (SMR) technology, the projected HDD capacity roadmap toward 50TB drives by 2030, and the potential role of SSDs in addressing storage performance challenges at CERN and beyond.