Exam Bundle

This signal belongs to the adjacent case “Daily aggregates contain duplicate transaction IDs from retries,” whose remedy is Deduplicate by the stable transaction identifier before summing amounts. The observed result for the present requirement is The output contains both dimensional subtotals and overall total.

Each account-region-month has the correct gross total pertains to a different control (Use `GROUP BY account_id, region, month` with `SUM(revenue)`) and a different need (The pipeline must aggregate monthly gross revenue by account and region). It cannot establish that Use supported `ROLLUP` or grouping sets with correct aggregate columns met the stated goal.

This signal belongs to the adjacent case “Null monetary amounts must not silently become fabricated transactions,” whose remedy is Treat nulls under the documented business policy before aggregation. The observed result for the present requirement is The output contains both dimensional subtotals and overall total.

Choose the newest event per key before merging is appropriate when “One input batch contains multiple changes per key and MERGE has ambiguous matches”; it would be verified by The MERGE source has at most one relevant row per target key. It does not directly satisfy “Matched keys should update and unseen keys should be inserted atomically,” whose relevant acceptance evidence is One committed transaction applies the intended upsert.

This action solves a different issue: “The full snapshot is authoritative and stale target keys must be purged.” Its normal signal is Only keys absent from the authorized snapshot scope are removed. Here the requested behavior is “Matched keys should update and unseen keys should be inserted atomically,” so that signal is not sufficient.

Use deterministic stable keys and idempotent matched/not-matched logic is appropriate when “A retried merge must not produce duplicate logical business records”; it would be verified by Re-execution leaves one current target row per key. It does not directly satisfy “Matched keys should update and unseen keys should be inserted atomically,” whose relevant acceptance evidence is One committed transaction applies the intended upsert.

Older tombstones cannot remove newer target versions would demonstrate the separate action Use a sequenced matched-delete predicate in MERGE, which addresses “Rows marked deleted should be removed only when the incoming event is newer.” It does not verify the requirement “Matched keys should update and unseen keys should be inserted atomically” following Use Delta `MERGE INTO` with matched update and not-matched insert.

This signal belongs to the adjacent case “A retried merge must not produce duplicate logical business records,” whose remedy is Use deterministic stable keys and idempotent matched/not-matched logic. The observed result for the present requirement is One committed transaction applies the intended upsert.

This signal belongs to the adjacent case “One input batch contains multiple changes per key and MERGE has ambiguous matches,” whose remedy is Choose the newest event per key before merging. The observed result for the present requirement is One committed transaction applies the intended upsert.

Apply a scoped `WHEN NOT MATCHED BY SOURCE THEN DELETE` clause is appropriate when “The full snapshot is authoritative and stale target keys must be purged”; it would be verified by Only keys absent from the authorized snapshot scope are removed. It does not directly satisfy “Rows marked deleted should be removed only when the incoming event is newer,” whose relevant acceptance evidence is Older tombstones cannot remove newer target versions.

The alternative Choose the newest event per key before merging has a distinct purpose—One input batch contains multiple changes per key and MERGE has ambiguous matches. It would not establish the required outcome (Rows marked deleted should be removed only when the incoming event is newer); the necessary verification concerns Older tombstones cannot remove newer target versions.

The alternative Use deterministic stable keys and idempotent matched/not-matched logic has a distinct purpose—A retried merge must not produce duplicate logical business records. It would not establish the required outcome (Rows marked deleted should be removed only when the incoming event is newer); the necessary verification concerns Older tombstones cannot remove newer target versions.

One committed transaction applies the intended upsert pertains to a different control (Use Delta `MERGE INTO` with matched update and not-matched insert) and a different need (Matched keys should update and unseen keys should be inserted atomically). It cannot establish that Use a sequenced matched-delete predicate in MERGE met the stated goal.

This signal belongs to the adjacent case “The full snapshot is authoritative and stale target keys must be purged,” whose remedy is Apply a scoped `WHEN NOT MATCHED BY SOURCE THEN DELETE` clause. The observed result for the present requirement is Older tombstones cannot remove newer target versions.

Re-execution leaves one current target row per key would demonstrate the separate action Use deterministic stable keys and idempotent matched/not-matched logic, which addresses “A retried merge must not produce duplicate logical business records.” It does not verify the requirement “Rows marked deleted should be removed only when the incoming event is newer” following Use a sequenced matched-delete predicate in MERGE.

This action solves a different issue: “The full snapshot is authoritative and stale target keys must be purged.” Its normal signal is Only keys absent from the authorized snapshot scope are removed. Here the requested behavior is “One input batch contains multiple changes per key and MERGE has ambiguous matches,” so that signal is not sufficient.

This action solves a different issue: “A retried merge must not produce duplicate logical business records.” Its normal signal is Re-execution leaves one current target row per key. Here the requested behavior is “One input batch contains multiple changes per key and MERGE has ambiguous matches,” so that signal is not sufficient.

The alternative Use Delta `MERGE INTO` with matched update and not-matched insert has a distinct purpose—Matched keys should update and unseen keys should be inserted atomically. It would not establish the required outcome (One input batch contains multiple changes per key and MERGE has ambiguous matches); the necessary verification concerns The MERGE source has at most one relevant row per target key.

That observation is a valid indicator for “A retried merge must not produce duplicate logical business records” when using Use deterministic stable keys and idempotent matched/not-matched logic. The scenario instead tests “One input batch contains multiple changes per key and MERGE has ambiguous matches”; the relevant result is The MERGE source has at most one relevant row per target key.

This signal belongs to the adjacent case “Matched keys should update and unseen keys should be inserted atomically,” whose remedy is Use Delta `MERGE INTO` with matched update and not-matched insert. The observed result for the present requirement is The MERGE source has at most one relevant row per target key.

That observation is a valid indicator for “The full snapshot is authoritative and stale target keys must be purged” when using Apply a scoped `WHEN NOT MATCHED BY SOURCE THEN DELETE` clause. The scenario instead tests “One input batch contains multiple changes per key and MERGE has ambiguous matches”; the relevant result is The MERGE source has at most one relevant row per target key.

The alternative Use Delta `MERGE INTO` with matched update and not-matched insert has a distinct purpose—Matched keys should update and unseen keys should be inserted atomically. It would not establish the required outcome (The full snapshot is authoritative and stale target keys must be purged); the necessary verification concerns Only keys absent from the authorized snapshot scope are removed.

Choose the newest event per key before merging belongs to the separate case “One input batch contains multiple changes per key and MERGE has ambiguous matches,” where The MERGE source has at most one relevant row per target key would be expected. That differs from the condition in this stem: “The full snapshot is authoritative and stale target keys must be purged.”

Use deterministic stable keys and idempotent matched/not-matched logic belongs to the separate case “A retried merge must not produce duplicate logical business records,” where Re-execution leaves one current target row per key would be expected. That differs from the condition in this stem: “The full snapshot is authoritative and stale target keys must be purged.”

That observation is a valid indicator for “Rows marked deleted should be removed only when the incoming event is newer” when using Use a sequenced matched-delete predicate in MERGE. The scenario instead tests “The full snapshot is authoritative and stale target keys must be purged”; the relevant result is Only keys absent from the authorized snapshot scope are removed.

Re-execution leaves one current target row per key pertains to a different control (Use deterministic stable keys and idempotent matched/not-matched logic) and a different need (A retried merge must not produce duplicate logical business records). It cannot establish that Apply a scoped `WHEN NOT MATCHED BY SOURCE THEN DELETE` clause met the stated goal.

The MERGE source has at most one relevant row per target key would demonstrate the separate action Choose the newest event per key before merging, which addresses “One input batch contains multiple changes per key and MERGE has ambiguous matches.” It does not verify the requirement “The full snapshot is authoritative and stale target keys must be purged” following Apply a scoped `WHEN NOT MATCHED BY SOURCE THEN DELETE` clause.

This action solves a different issue: “The full snapshot is authoritative and stale target keys must be purged.” Its normal signal is Only keys absent from the authorized snapshot scope are removed. Here the requested behavior is “A retried merge must not produce duplicate logical business records,” so that signal is not sufficient.

Choose the newest event per key before merging belongs to the separate case “One input batch contains multiple changes per key and MERGE has ambiguous matches,” where The MERGE source has at most one relevant row per target key would be expected. That differs from the condition in this stem: “A retried merge must not produce duplicate logical business records.”

Use a sequenced matched-delete predicate in MERGE is appropriate when “Rows marked deleted should be removed only when the incoming event is newer”; it would be verified by Older tombstones cannot remove newer target versions. It does not directly satisfy “A retried merge must not produce duplicate logical business records,” whose relevant acceptance evidence is Re-execution leaves one current target row per key.

Only keys absent from the authorized snapshot scope are removed would demonstrate the separate action Apply a scoped `WHEN NOT MATCHED BY SOURCE THEN DELETE` clause, which addresses “The full snapshot is authoritative and stale target keys must be purged.” It does not verify the requirement “A retried merge must not produce duplicate logical business records” following Use deterministic stable keys and idempotent matched/not-matched logic.

That observation is a valid indicator for “Rows marked deleted should be removed only when the incoming event is newer” when using Use a sequenced matched-delete predicate in MERGE. The scenario instead tests “A retried merge must not produce duplicate logical business records”; the relevant result is Re-execution leaves one current target row per key.

One committed transaction applies the intended upsert would demonstrate the separate action Use Delta `MERGE INTO` with matched update and not-matched insert, which addresses “Matched keys should update and unseen keys should be inserted atomically.” It does not verify the requirement “A retried merge must not produce duplicate logical business records” following Use deterministic stable keys and idempotent matched/not-matched logic.