DAMON Subsystem Adds Tiering and THP Monitoring in Major 2026 Update

Breaking: DAMON Subsystem Unveils Major Expansion at Linux Storage Summit 2026

VANCOUVER, BC — The kernel's DAMON subsystem, a critical tool for memory monitoring and management, is undergoing a significant expansion with new capabilities including memory tiering, data attributes monitoring, and transparent huge pages (THP) support. The update was presented by DAMON creator SeongJae Park at the 2026 Linux Storage, Filesystem, Memory Management, and BPF Summit, continuing a tradition of annual progress reports for the rapidly evolving subsystem.

"DAMON has become indispensable for advanced memory management in Linux," said Park during his presentation. "This release introduces features that allow systems to dynamically adapt to memory pressure and workload patterns, moving us closer to fully autonomous memory optimization."

Background

DAMON (Data Access MONitor) was initially designed to provide user-space visibility into system memory access patterns. It helps administrators and automated tools make intelligent decisions about memory allocation, reclaim, and placement. Since its introduction, DAMON has seen rapid development, with yearly updates becoming a staple of the Linux Storage Summit.

Prior to this update, DAMON offered basic monitoring of memory access frequencies and regions. The subsystem operates as a lightweight, online framework that can be used by higher-level memory management policies without incurring significant overhead.

New Capabilities in Detail

Among the most notable additions is memory tiering support. This feature enables systems to classify memory into performance tiers (e.g., DRAM, persistent memory, CXL-attached memory) and automatically migrate data between them based on access patterns. Park highlighted that tiering is especially relevant as heterogeneous memory architectures become commonplace in data centers and AI workloads.

Another key addition is data attributes monitoring. While DAMON previously tracked only access frequency, it now can monitor attributes such as read vs. write intensity, temporal locality, and reuse distance. This richer dataset allows for more nuanced management decisions, such as selectively promoting write-heavy pages to higher tiers.

"With data attributes, we can differentiate between hot data that is mostly read and hot data that is being constantly modified," explained Park. "This granularity unlocks new optimization strategies for file systems and memory managers."

Additionally, transparent huge pages (THP) support has been integrated. DAMON can now monitor THP usage and fragmentation, providing feedback to the kernel's THP subsystem to improve performance. This includes detecting when huge pages cause memory waste or when splitting them could relieve pressure.

What This Means

For Linux system administrators and kernel developers, this update represents a leap forward in memory observability and control. The combination of tiering and attribute monitoring enables more precise memory policies without requiring manual tuning. Over time, this could lead to self-optimizing Linux systems that automatically adapt to changing workloads.

The expanded monitoring also has implications for cloud and hyperscale environments, where memory costs dominate operational expenses. By efficiently placing data in the right tier and splitting huge pages only when necessary, organizations can reduce memory footprint while maintaining performance. Park noted that early internal tests showed up to 15% memory savings in mixed workloads.

"This is not just a minor feature release — it's a foundation for the next generation of memory management in Linux," said Park. "We are enabling systems to understand memory in much richer detail than ever before."

Developers interested in trying these new features can find the patches in the DAMON mailing list. The updates are expected to be merged into the mainline kernel during the 2026 merge window, with full documentation following shortly thereafter.