This landscape covers patent families whose title or abstract references a storage engine, LSM tree or columnar storage, combined with claim-level language on write amplification, compaction, compression ratio, point lookup or scan performance. It is restricted to IPC classes G06F16, G06F12 and G06F3, which cover data retrieval, memory architecture and input/output arrangements respectively. That combination excludes general database-query patents that don't touch the underlying storage layer, and excludes storage-hardware patents that don't address data layout or compaction logic.

No - filings rose steadily from 7 in 2017 to a peak of 50 in 2023, but the midpoint year of 2022 was already at 42, meaning most of the growth had happened before the peak. The trend has since flattened rather than continued climbing. Because publication lags filing by roughly 18 months, the 2025 and 2026 figures will revise upward somewhat, but the broader pattern is a plateau, not sustained growth.

China leads by receiving-office count, with 161 records against 86 from the United States in this dataset. Europe (EPO), WIPO/PCT filings, Hong Kong and South Korea make up the remainder in much smaller numbers. For anyone assessing where competitive claim density is highest, a China-first search of assignees and claim language is essential rather than optional.

US11237744B2, assigned to Yahoo Assets LLC and granted in 2022, claims a method for deriving a storage engine's write amplification factor from a compaction value - the minimum count of valid records in a data file needed to trigger clean-up. It ties a specific configuration parameter (compaction value) to a specific performance metric (write amplification) through a counter-and-threshold mechanism. Designs that determine write amplification independently of a stored compaction value, or that trigger compaction through different criteria such as fixed time intervals or size thresholds unrelated to record counts, sit outside its literal claim scope, but any implementation using this exact linkage should be reviewed against it directly.

The thinnest filing density relative to the core LSM-tree and columnar-storage claims appears in adaptive compaction scheduling that responds to mixed read/write load, tiered compression ratio selection based on hot/cold data classification, and scan-performance optimization for hybrid row-columnar layouts. These are adjacent to heavily-claimed core mechanics but are not yet occupied by dense prior art in this corpus. That doesn't guarantee novelty for any specific claim, but it does mark areas where a full clearance search is more likely to come back clean than in core compaction or LSM indexing claims.