Installing Fedora 45 on Stratis storage

A new option, Stratis, is introduced in the Anaconda installer for Fedora 45, offering a modern method for local storage management by integrating existing technologies like device mapper, LUKS, and XFS. Stratis operates on two core concepts: pools, which abstract block devices, and filesystems, which reside on those pools and serve as volumes. Pools can combine multiple devices and support features like encryption or over-provisioning. The installer provides an option to use a web-based user interface via the Cockpit storage module for manual configuration, which is necessary if not using automated options. Users must manually create partitions for booting before creating Stratis pools, and then subsequently define filesystems on those pools. Advanced users can utilize full kickstart support with custom commands like stratispool and stratisfs for automated setup involving pools and filesystems.

The introduction of Stratis bridges the gap between complex, layered storage technologies and user accessibility by abstracting complexity behind familiar tools. The pattern here suggests a systemic move toward simplifying infrastructure management, leveraging existing, proven components rather than introducing entirely novel paradigms. The dual pathways—graphical configuration via Cockpit and advanced scripting via kickstart—reflect an awareness that usability requires accommodating both novice installers and power users simultaneously. However, the tension lies in how readily complex concepts like over-provisioning can be managed; while automation is offered, the necessary cognitive load for understanding capacity versus allocation remains a critical point of friction. The system shifts control from the user directly interfacing with raw block devices to managing abstract pools, which requires a high degree of trust in the abstraction layer. The implication for agency is that while configuration becomes easier, the responsibility to understand the trade-offs inherent in pooling and provisioning must be consciously managed, lest the perceived simplicity lead to unforeseen resource exhaustion or security exposure when features like encryption are toggled post-installation. What further structures must be established to ensure that these layers of abstraction do not become new barriers to understanding data sovereignty?