Plan değişimi P586

(@incTest bit,@types1 nvarchar(4000),@sid int,@dbName nvarchar(4000),@model nvarchar(4000),@q vector(1024),@top int)SELECT TOP (@top) c.ChunkId, c.DocId, d.SourceType, d.SourceRef, d.Title, c.Heading, c.Text,
       VECTOR_DISTANCE('cosine', c.Embedding, @q) AS Score
FROM rag.Chunk AS c
JOIN rag.Document AS d ON d.DocId = c.DocId
WHERE (d.IsActive = 1 OR (@incTest = 1 AND d.ExclusionReason = 'TestDb')) AND d.SourceType IN (@types1)
AND (@sid IS NULL OR d.ServerID = @sid) AND (@dbName IS NULL OR d.DatabaseName = @dbName) AND c.Embedding IS NOT NULL AND c.EmbeddingModel = @model
ORDER BY Score, c.ChunkId

Plan change case #P586 | DB: DB-11 | Regression | Confidence: Low (review pending)
Query: (@incTest bit,@types1 nvarchar(?),@sid int,@dbName nvarchar(?),@model nvarchar(?),@q vector(?),@top int)SELECT TOP (@top) c.ChunkId, c.DocId, d.SourceType, d.SourceRef, d.Title, c.Heading, c.Text, VECTOR_DISTANCE(?, c.Embedding, @q) AS Score FROM rag.Chunk AS c JOIN rag.Document AS d ON d.DocId = c.DocId WHERE (d.IsActive = ? OR (@incTest = ? AND d.ExclusionReason = ?)) AND d.SourceType IN (@types1) AND (@sid IS NULL OR d.ServerID = @sid) AND (@dbName IS NULL OR d.DatabaseName = @dbName) AND c.Embedding IS NOT NULL AND c.EmbeddingModel = @model ORDER BY Score, c.ChunkId
Change: 2026-10-06 06:00 UTC, plan 62531 → 64705, pattern: Transition
Impact (Query Store, per execution): duration 3.6 → 39.8 ms (+1015%); CPU 3.6 → 29.1 ms (+715%); reads 211 → 6,170 (+2824%); rows 1.17 → 20 (+1614%)
Plan difference: Access only in the old plan: rag.Document Index Scan [UQ_Document_Source]; rag.Document Clustered Index Seek [PK_Document] (lookup); rag.Chunk Index Seek [UQ_Chunk_DocSeq]; rag.Chunk Clustered Index Seek [PK_Chunk] (lookup); Access only in the new plan: rag.Document Clustered Index Scan [PK_Document]; rag.Chunk Clustered Index Scan [PK_Chunk]; Join only in the old plan: Nested Loops (Inner Join) ×3; Join only in the new plan: Hash Match (Inner Join)
Candidate cause: STATS — Statistic rag.Chunk.PK_Chunk was updated 2026-10-05 22:47 (sampled 47.4%); 4 updates in the window; the plans use this statistic.
Candidate cause: STATS — Statistic rag.Chunk.UQ_Chunk_DocSeq was updated 2026-10-05 22:47 (sampled 50.7%); 4 updates in the window; the plans use this statistic.
Candidate cause: STATS — Statistic rag.Chunk._WA_Sys_00000007_49309088 was updated 2026-10-05 22:47 (sampled 50.5%); 3 updates in the window; the plans use this statistic.
Candidate cause: STATS — Statistic rag.Chunk._WA_Sys_00000003_49309088 was updated 2026-10-05 22:47 (sampled 47.4%); 4 updates in the window; the plans use this statistic.
Candidate cause: STATS — Statistic rag.Document.PK_Document was updated 2026-10-06 01:16 (sampled 100%); 2 updates in the window; the plans use this statistic.
Candidate cause: STATS — Statistic rag.Document.UQ_Document_Source was updated 2026-10-06 04:00 (sampled 100%); 3 updates in the window; the plans use this statistic.
Candidate cause: STATS — Statistic rag.Document._WA_Sys_0000000A_428392F9 was updated 2026-10-05 22:57 (sampled 100%); 3 updates in the window; the plans use this statistic.
Candidate cause: STATS — Statistic rag.Document._WA_Sys_00000005_428392F9 was updated 2026-10-05 22:46 (sampled 100%); 3 updates in the window; the plans use this statistic.
Candidate cause: STATS — Statistic rag.Document._WA_Sys_0000000E_428392F9 was updated 2026-10-06 05:48 (sampled 100%); 3 updates in the window; the plans use this statistic.
Candidate cause: WORKLOAD — The plans see a different workload: rows per execution 1.17 → 20. The plans may have been compiled for different parameters or batch sizes.
Counter-evidence: WORK_PROPORTIONAL — Rows per execution changed 17.14× and duration 11.15×; duration per row 3.0598 → 1.9894 ms: the duration difference comes from the amount of data processed, not the plan.
Counter-evidence: HIGH_VARIANCE — The new plan's duration standard deviation is 263% of the average: the average is not reliable.
Confidence rationale: Low: strong counter-evidence (WORK_PROPORTIONAL).
Note: this is a candidate cause, not proof of the root cause.