Which Cosmos DB consistency level avoids out-of-order writes in multi-region reads?
Short answer
D Consistent prefix ensures multi-region reads never receive out-of-order writes while maximizing read concurrency.
Question
Key Takeaways
Tests which Cosmos DB consistency level preserves write order without an unnecessary staleness bound; the trap is choosing bounded staleness because it also orders reads but is stronger than required.
Azure Cosmos DB consistent prefix is the weakest consistency level that guarantees reads never see out-of-order writes while maximizing multi-region read concurrency. This AZ-204 guide confirms option D as the correct consistency setting for the stated Cosmos DB requirements.
Common mistake: Many candidates pick bounded staleness (C) because it also prevents out-of-order reads, but it imposes a staleness window and therefore does not maximize read concurrency as fully as consistent prefix.
Expert Analysis
Why the Answer Is Correct
Consistent prefix consistency in Azure Cosmos DB guarantees that read operations never see writes out of order; updates are observed in the same sequence they were committed. The question asks only for ordering, not for a maximum lag or version bound, so the weakest consistency level that still guarantees ordering is the best fit. Because consistent prefix does not enforce a staleness window, it leaves more read concurrency and availability available across all Azure regions. Therefore option D satisfies both the no-out-of-order-read requirement and the maximize-concurrency requirement.Why the Other Options Are Wrong
Session consistency (A) provides monotonic reads, read-your-writes, and write-follows-reads only within a single client session, not globally across all regions. Eventual consistency (B) allows reads to see writes out of order, which directly violates the first requirement. Bounded staleness (C) also preserves consistent-prefix ordering, but it adds a guaranteed staleness bound of K versions or T time, making it stronger than necessary and reducing read concurrency relative to consistent prefix. Strong consistency would likewise meet ordering but would further reduce availability and concurrency, confirming that consistent prefix is the intended weakest ordering-preserving level.Community Comment Notes
As de55d65 noted, eventual and bounded staleness can be ruled out because eventual permits out-of-order reads and the question provides no staleness boundary requirement; the lower the consistency, the higher the concurrency, so D fits. Mattt and Vichu_1607 argued for bounded staleness (C), but bounded staleness adds an explicit lag window that the scenario does not request and that is stronger than consistent prefix. passme847 quoted the Microsoft documentation: “the reads never see out of order writes” for consistent prefix, which directly matches the requirement. Jay456 agreed with de55d65 after reviewing the documentation, while ns4098 asked “Why not C?” — the answer is that C is stronger than the stated ordering need.Official Reference
https://learn.microsoft.com/en-us/azure/cosmos-db/consistency-levels
This link opens the official vendor documentation in a new tab.
Exam Strategy
Focus on the weakest consistency level that still meets every stated guarantee; “maximize concurrency” is the signal to avoid stronger models like bounded staleness or strong. Read the requirement keywords literally: out-of-order protection points to consistent prefix.
Frequently Asked Questions
Why isn't bounded staleness the best choice for multi-region Cosmos DB reads?
Does session consistency prevent out-of-order reads across all Azure regions?
Community Votes
57% of anonymous learners picked answer D. Votes are pick records left by other test-takers — they are not the verified answer.
Discussion 7 contributions
Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.
From public community threads 7
-
de55d65 👍 8 Answered DWhat we can rule out is B eventual, since there can be out-of-order reads, and C, since the question does not contain any information about a boundary. If the wording of the question was "consistency" instead of "concurrency", I would go for A "session". Therefore I'd expect the lower the consistency, the higher the concurrency. I would go for D Consistent prefix. Reference: https://learn.microsoft.com/en-us/azure/cosmos-db/consistency-levels No words about concurrency there.
-
Mattt 👍 4 Answered CC is the correct answer Bounded staleness guarantees that read operations will never see writes out of order, which directly addresses your requirement for ensuring that read operations never receive write operations that are out of order. It also maximizes the concurrency of read operations in multiple regions while providing a guarantee that reads lag behind writes by a certain staleness window (defined in terms of time or number of versions). This balance between consistency and availability fits your scenario well.
-
Vichu_1607 👍 3 Answered CC. bounded staleness: Bounded staleness consistency ensures that reads never see out-of-order writes and provides a lag window within which the data might be stale. This consistency level guarantees that reads are consistent within a specified staleness window, which meets the requirement of never receiving out-of-order writes. It also allows for high concurrency of read operations across multiple regions.
-
Jay456 👍 2 Answered DAfter reading the documentation I fully agree with de55d65's explanation and choose D as well.
-
passme847 👍 1 Answered D"Consistent prefix consistency, the reads never see out of order writes for a transactional batch of writes" https://learn.microsoft.com/en-us/azure/cosmos-db/consistency-levels
Show 2 more comments
-
DanielRO 👍 1 Answered DChatGPT response: Consistent prefix.
-
ns4098 👍 2 Answered CWhy not C?
Reader comments & corrections
No reader corrections yet — spotted an error or have a note? Share it below.