Understanding Inconsistency in Azure Cosmos DB with TLA+
Bibliographic record
Abstract
Beyond implementation correctness of a distributed system, it is equally important to understand exactly what users should expect to see from that system. Even if the system itself works as designed, insufficient understanding of its user-observable semantics can cause bugs in its dependencies. By focusing a formal specification effort on precisely defining the expected user-observable behaviors of the Azure Cosmos DB service at Microsoft, we were able to write a formal specification of the database that was significantly smaller and conceptually simpler than any other specification of Cosmos DB, while representing a wider range of valid user-observable behaviors than existing more detailed specifications. Many of the additional behaviors we documented were previously poorly understood outside of the Cosmos DB development team, even informally, leading to data consistency errors in Microsoft products that depend on it. Using this specification, we were able to raise two key issues in Cosmos DB’s public-facing documentation, which have since been addressed. We were also able to offer a fundamental solution to a previous high-impact outage within another Azure service that depends on Cosmos DB.
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 | 0.021 | 0.044 |
| Meta-epidemiology (narrow) | 0.001 | 0.002 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.002 | 0.004 |
| Scholarly communication | 0.007 | 0.010 |
| Open science | 0.004 | 0.005 |
| Research integrity | 0.002 | 0.005 |
| Insufficient payload (model declined to judge) | 0.005 | 0.001 |
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".