Wayfinder: Speeding up Key-Value Separation by Avoiding I/O Based Indirection
Bibliographic record
Abstract
LSM-Trees are the backbone of modern key-value stores, supporting write-intensive workloads with balanced performance for point and range queries. Compaction in LSM-trees optimizes queries but poses significant overhead on the write path, especially for medium to large values. Key-value (KV) separation addresses this by storing values in a separate value log and pointing to them from the LSM-tree. This reduces write amplification as values are no longer rewritten during compaction. This KV separation, however, presents its own challenges. First, query performance suffers as the LSM-tree must first be searched for an address followed by querying the log for the associated value. Second, the value log requires expensive garbage collection as (1) the LSM-tree must be queried to determine whether a given value in the log is still the most recent version, and (2) a reinsertion is needed to update the LSM-tree with the addresses of the relocated KV pairs. To address these challenges, this paper investigates how to map values in the log via an in-memory hash table while using the LSM-tree to store small values and handle range queries. We implement Wayfinder on top of RocksDB and show its effectiveness in improving throughput while reducing space and write amplification.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.011 |
| Open science | 0.004 | 0.006 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.010 | 0.004 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".