statistics: add default value of CMSketch for Analyze (#19455) (#19927

)

Signed-off-by: ti-srebot <[email protected]>

executor/analyze.go

@@ -335,6 +335,9 @@ func (e *AnalyzeIndexExec) buildStatsFromResult(result distsql.SelectResult, nee
 		}
 	}
 	err := hist.ExtractTopN(cms, len(e.idxInfo.Columns), uint32(e.opts[ast.AnalyzeOptNumTopN]))
+	if needCMS && cms != nil {
+		cms.CalcDefaultValForAnalyze(uint64(hist.NDV))
+	}
 	return hist, cms, err
 }
 
@@ -525,6 +528,7 @@ func (e *AnalyzeColumnsExec) buildStats(ranges []*ranger.Range) (hists []*statis
 			return nil, nil, err
 		}
 		hists = append(hists, hg)
+		collectors[i].CMSketch.CalcDefaultValForAnalyze(uint64(hg.NDV))
 		cms = append(cms, collectors[i].CMSketch)
 	}
 	return hists, cms, nil
@@ -1236,6 +1240,7 @@ func analyzeIndexIncremental(idxExec *analyzeIndexIncrementalExec) analyzeResult
 		if err != nil {
 			return analyzeResult{Err: err, job: idxExec.job}
 		}
+		cms.CalcDefaultValForAnalyze(uint64(hist.NDV))
 	}
 	result := analyzeResult{
 		PhysicalTableID: idxExec.physicalTableID,

executor/analyze_test.go

@@ -592,3 +592,35 @@ func (s *testSuite1) TestHashInTopN(c *C) {
 		}
 	}
 }
+
+func (s *testSuite1) TestDefaultValForAnalyze(c *C) {
+	tk := testkit.NewTestKit(c, s.store)
+	tk.MustExec("drop database if exists test_default_val_for_analyze;")
+	tk.MustExec("create database test_default_val_for_analyze;")
+	tk.MustExec("use test_default_val_for_analyze")
+
+	tk.MustExec("create table t (a int, key(a));")
+	for i := 0; i < 2048; i++ {
+		tk.MustExec("insert into t values (0)")
+	}
+	for i := 1; i < 4; i++ {
+		tk.MustExec("insert into t values (?)", i)
+	}
+	tk.MustExec("analyze table t with 0 topn;")
+	tk.MustQuery("explain select * from t where a = 1").Check(testkit.Rows("IndexReader_6 512.00 root  index:IndexRangeScan_5",
+		"└─IndexRangeScan_5 512.00 cop[tikv] table:t, index:a(a) range:[1,1], keep order:false"))
+	tk.MustQuery("explain select * from t where a = 999").Check(testkit.Rows("IndexReader_6 0.00 root  index:IndexRangeScan_5",
+		"└─IndexRangeScan_5 0.00 cop[tikv] table:t, index:a(a) range:[999,999], keep order:false"))
+
+	tk.MustExec("drop table t;")
+	tk.MustExec("create table t (a int, key(a));")
+	for i := 0; i < 2048; i++ {
+		tk.MustExec("insert into t values (0)")
+	}
+	for i := 1; i < 2049; i++ {
+		tk.MustExec("insert into t values (?)", i)
+	}
+	tk.MustExec("analyze table t with 0 topn;")
+	tk.MustQuery("explain select * from t where a = 1").Check(testkit.Rows("IndexReader_6 1.00 root  index:IndexRangeScan_5",
+		"└─IndexRangeScan_5 1.00 cop[tikv] table:t, index:a(a) range:[1,1], keep order:false"))
+}

statistics/cmsketch.go

@@ -288,23 +288,32 @@ func (c *CMSketch) QueryBytes(d []byte) uint64 {
 func (c *CMSketch) queryHashValue(h1, h2 uint64) uint64 {
 	vals := make([]uint32, c.depth)
 	min := uint32(math.MaxUint32)
+	// We want that when res is 0 before the noise is eliminated, the default value is not used.
+	// So we need a temp value to distinguish before and after eliminating noise.
+	temp := uint32(1)
 	for i := range c.table {
 		j := (h1 + h2*uint64(i)) % uint64(c.width)
 		if min > c.table[i][j] {
 			min = c.table[i][j]
 		}
 		noise := (c.count - uint64(c.table[i][j])) / (uint64(c.width) - 1)
-		if uint64(c.table[i][j]) < noise {
+		if uint64(c.table[i][j]) == 0 {
 			vals[i] = 0
+		} else if uint64(c.table[i][j]) < noise {
+			vals[i] = temp
 		} else {
-			vals[i] = c.table[i][j] - uint32(noise)
+			vals[i] = c.table[i][j] - uint32(noise) + temp
 		}
 	}
 	sort.Sort(sortutil.Uint32Slice(vals))
 	res := vals[(c.depth-1)/2] + (vals[c.depth/2]-vals[(c.depth-1)/2])/2
-	if res > min {
-		res = min
+	if res > min+temp {
+		res = min + temp
 	}
+	if res == 0 {
+		return uint64(0)
+	}
+	res = res - temp
 	if c.considerDefVal(uint64(res)) {
 		return c.defaultValue
 	}
@@ -538,3 +547,16 @@ func (c *CMSketch) AppendTopN(data []byte, count uint64) {
 func (c *CMSketch) GetWidthAndDepth() (int32, int32) {
 	return c.width, c.depth
 }
+
+// CalcDefaultValForAnalyze calculate the default value for Analyze.
+// The value of it is count / NDV in CMSketch. This means count and NDV are not include topN.
+func (c *CMSketch) CalcDefaultValForAnalyze(NDV uint64) {
+	// If NDV <= TopN, all values should be in TopN.
+	// So we set c.defaultValue to 0 and return immediately.
+	if NDV <= uint64(len(c.topN)) {
+		c.defaultValue = 0
+		return
+	}
+	remainNDV := NDV - uint64(len(c.topN))
+	c.defaultValue = c.count / mathutil.MaxUint64(1, remainNDV)
+}