Ejemplo n.º 1
0
        private WBooleanExpression ConstructJoinCondition(
            CandidateJoinUnit candidateTree,
            IMatchJoinStatisticsCalculator statisticsCalculator,
            GraphMetaData metaData,
            out double joinSelectivity,
            out double sqlEstimatedJoinSelectivity)
        {
            joinSelectivity             = 1.0;
            sqlEstimatedJoinSelectivity = 1.0;


            var root = candidateTree.TreeRoot;

            WBooleanExpression joinCondition = null;
            string             nodeName      = "";


            // Update Nodes
            if (MaterializedNodeSplitCount.ContainsKey(root))
            {
                MaterializedNodeSplitCount[root]++;
                nodeName      = GetNodeRefName(root);
                joinCondition = new WBooleanComparisonExpression
                {
                    FirstExpr = new WColumnReferenceExpression
                    {
                        ColumnType          = ColumnType.Regular,
                        MultiPartIdentifier = new WMultiPartIdentifier(
                            new Identifier {
                            Value = root.RefAlias
                        },
                            new Identifier {
                            Value = "GlobalNodeId"
                        }
                            ),
                    },
                    SecondExpr = new WColumnReferenceExpression
                    {
                        ColumnType          = ColumnType.Regular,
                        MultiPartIdentifier = new WMultiPartIdentifier(
                            new Identifier {
                            Value = nodeName
                        },
                            new Identifier {
                            Value = "GlobalNodeId"
                        }
                            ),
                    },
                    ComparisonType = BooleanComparisonType.Equals
                };
            }
            else
            {
                nodeName = root.RefAlias;
                if (!Nodes.Contains(root))
                {
                    Nodes.Add(root);
                }
                MaterializedNodeSplitCount[root] = 0;
            }

            List <double> densityList = new List <double>();

            List <MatchEdge> inEdges;

            if (UnmaterializedNodeMapping.TryGetValue(root, out inEdges))
            {
                var  firstEdge    = inEdges.First();
                bool materialized = EdgeMaterilizedDict[firstEdge];
                UnmaterializedNodeMapping.Remove(root);
                joinSelectivity *= 1.0 / root.TableRowCount;

                // Component materialized edge to root
                if (materialized)
                {
                    joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition, new WBooleanComparisonExpression
                    {
                        FirstExpr = new WColumnReferenceExpression
                        {
                            ColumnType          = ColumnType.Regular,
                            MultiPartIdentifier = new WMultiPartIdentifier(
                                new Identifier {
                                Value = firstEdge.EdgeAlias
                            },
                                new Identifier {
                                Value = "Sink"
                            }
                                ),
                        },
                        SecondExpr = new WColumnReferenceExpression
                        {
                            ColumnType          = ColumnType.Regular,
                            MultiPartIdentifier = new WMultiPartIdentifier(
                                new Identifier {
                                Value = nodeName
                            },
                                new Identifier {
                                Value = "GlobalNodeId"
                            }
                                )
                        },
                        ComparisonType = BooleanComparisonType.Equals
                    });

                    densityList.Add(root.GlobalNodeIdDensity);
                }
                // Component unmaterialized edge to root
                else
                {
                    Statistics statistics = null;
                    foreach (var edge in inEdges)
                    {
                        // Update component table
                        TableRef = SpanTableRef(TableRef, edge, GetNodeRefName(edge.SourceNode), metaData);

                        EdgeMaterilizedDict[edge] = true;
                        joinCondition             = WBooleanBinaryExpression.Conjunction(joinCondition,
                                                                                         new WBooleanComparisonExpression
                        {
                            FirstExpr = new WColumnReferenceExpression
                            {
                                ColumnType          = ColumnType.Regular,
                                MultiPartIdentifier = new WMultiPartIdentifier(
                                    new Identifier {
                                    Value = edge.EdgeAlias
                                },
                                    new Identifier {
                                    Value = "Sink"
                                }
                                    ),
                            },
                            SecondExpr = new WColumnReferenceExpression
                            {
                                ColumnType          = ColumnType.Regular,
                                MultiPartIdentifier = new WMultiPartIdentifier(
                                    new Identifier {
                                    Value = nodeName
                                },
                                    new Identifier {
                                    Value = "GlobalNodeId"
                                }
                                    )
                            },
                            ComparisonType = BooleanComparisonType.Equals
                        });
                        double selectivity;
                        statistics = Statistics.UpdateHistogram(statistics,
                                                                edge.Statistics, out selectivity);
                        joinSelectivity *= selectivity;
                        densityList.Add(root.GlobalNodeIdDensity);
                    }
                    SinkNodeStatisticsDict[root] = statistics;
                }
            }

            var jointEdges = candidateTree.MaterializedEdges;

            foreach (var jointEdge in jointEdges)
            {
                EdgeMaterilizedDict[jointEdge] = true;
                var sinkNode = jointEdge.SinkNode;
                // Leaf to component materialized node
                if (MaterializedNodeSplitCount.ContainsKey(sinkNode))
                {
                    joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition,
                                                                         new WBooleanComparisonExpression
                    {
                        FirstExpr = new WColumnReferenceExpression
                        {
                            ColumnType          = ColumnType.Regular,
                            MultiPartIdentifier = new WMultiPartIdentifier(
                                new Identifier {
                                Value = jointEdge.EdgeAlias
                            },
                                new Identifier {
                                Value = "Sink"
                            }
                                ),
                        },
                        SecondExpr = new WColumnReferenceExpression
                        {
                            ColumnType          = ColumnType.Regular,
                            MultiPartIdentifier = new WMultiPartIdentifier(
                                new Identifier {
                                Value = sinkNode.RefAlias
                            },
                                new Identifier {
                                Value = "GlobalNodeId"
                            }
                                )
                        },
                        ComparisonType = BooleanComparisonType.Equals
                    });
                    Statistics sinkNodeStatistics;
                    if (!SinkNodeStatisticsDict.TryGetValue(sinkNode, out sinkNodeStatistics))
                    {
                        sinkNodeStatistics = null;
                        joinSelectivity   *= 1.0 / sinkNode.TableRowCount;
                    }
                    double selectivity;
                    var    statistics = Statistics.UpdateHistogram(sinkNodeStatistics,
                                                                   jointEdge.Statistics, out selectivity);
                    joinSelectivity *= selectivity;
                    SinkNodeStatisticsDict[sinkNode] = statistics;
                    densityList.Add(sinkNode.GlobalNodeIdDensity);
                }
                // Leaf to component unmaterialized node
                else
                {
                    inEdges = UnmaterializedNodeMapping[sinkNode];
                    var  firstEdge      = inEdges.First();
                    bool materlizedEdge = EdgeMaterilizedDict[firstEdge];

                    // Leaf to materialized leaf
                    if (materlizedEdge)
                    {
                        joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition,
                                                                             new WBooleanComparisonExpression
                        {
                            FirstExpr = new WColumnReferenceExpression
                            {
                                ColumnType          = ColumnType.Regular,
                                MultiPartIdentifier = new WMultiPartIdentifier(
                                    new Identifier {
                                    Value = jointEdge.EdgeAlias
                                },
                                    new Identifier {
                                    Value = "Sink"
                                }
                                    ),
                            },
                            SecondExpr = new WColumnReferenceExpression
                            {
                                ColumnType          = ColumnType.Regular,
                                MultiPartIdentifier = new WMultiPartIdentifier(
                                    new Identifier {
                                    Value = firstEdge.EdgeAlias
                                },
                                    new Identifier {
                                    Value = "Sink"
                                }
                                    )
                            },
                            ComparisonType = BooleanComparisonType.Equals
                        });

                        densityList.Add(Statistics.DefaultDensity);
                        double selectivity;
                        var    statistics = Statistics.UpdateHistogram(SinkNodeStatisticsDict[sinkNode],
                                                                       jointEdge.Statistics, out selectivity);
                        joinSelectivity *= selectivity;
                        SinkNodeStatisticsDict[sinkNode] = statistics;
                    }
                    // Leaf to unmaterialized leaf
                    else
                    {
                        Statistics compSinkNodeStatistics = null;
                        foreach (var inEdge in inEdges)
                        {
                            TableRef = SpanTableRef(TableRef, inEdge, GetNodeRefName(inEdge.SourceNode), metaData);
                            EdgeMaterilizedDict[inEdge] = true;
                            joinCondition = WBooleanBinaryExpression.Conjunction(joinCondition,
                                                                                 new WBooleanComparisonExpression
                            {
                                FirstExpr = new WColumnReferenceExpression
                                {
                                    ColumnType          = ColumnType.Regular,
                                    MultiPartIdentifier = new WMultiPartIdentifier(
                                        new Identifier {
                                        Value = jointEdge.EdgeAlias
                                    },
                                        new Identifier {
                                        Value = "Sink"
                                    }
                                        ),
                                },
                                SecondExpr = new WColumnReferenceExpression
                                {
                                    ColumnType          = ColumnType.Regular,
                                    MultiPartIdentifier = new WMultiPartIdentifier(
                                        new Identifier {
                                        Value = inEdge.EdgeAlias
                                    },
                                        new Identifier {
                                        Value = "Sink"
                                    }
                                        )
                                },
                                ComparisonType = BooleanComparisonType.Equals
                            });

                            densityList.Add(Statistics.DefaultDensity);

                            double selectivity;
                            var    leafToLeafStatistics = statisticsCalculator.GetLeafToLeafStatistics(jointEdge, inEdge,
                                                                                                       out selectivity);
                            joinSelectivity       *= selectivity;
                            compSinkNodeStatistics =
                                Statistics.UpdateHistogram(compSinkNodeStatistics,
                                                           inEdge.Statistics, out selectivity);
                        }
                        SinkNodeStatisticsDict[sinkNode] = compSinkNodeStatistics;
                    }
                }
            }

            var unmatEdges = candidateTree.UnmaterializedEdges;

            foreach (var unmatEdge in unmatEdges)
            {
                EdgeMaterilizedDict[unmatEdge] = false;
                if (!Nodes.Contains(unmatEdge.SinkNode))
                {
                    Nodes.Add(unmatEdge.SinkNode);
                }
                var sinkNodeInEdges = UnmaterializedNodeMapping.GetOrCreate(unmatEdge.SinkNode);
                sinkNodeInEdges.Add(unmatEdge);
            }

            // Calculate Estimated Join Selectivity & Estimated Node Size
            densityList.Sort();
            for (int i = densityList.Count - 1; i >= 0; i--)
            {
                sqlEstimatedJoinSelectivity *= Math.Sqrt(sqlEstimatedJoinSelectivity) * densityList[i];
            }

            return(joinCondition);
        }
Ejemplo n.º 2
0
        /// <summary>
        /// Calculate join costs and update components using optimal join method & order
        /// </summary>
        /// <param name="nodeUnitCandidate"></param>
        /// <param name="joinCondition"></param>
        /// <param name="joinSelectivity"></param>
        /// <param name="estimatedSelectivity"></param>
        /// <returns></returns>
        private void ConstructPhysicalJoinAndUpdateCost(
            CandidateJoinUnit nodeUnitCandidate,
            WBooleanExpression joinCondition,
            double joinSelectivity,
            double estimatedSelectivity,
            GraphMetaData metaData)
        {
            var nodeDegrees           = nodeUnitCandidate.EdgeDegrees;
            var nodeUnitSize          = nodeUnitCandidate.TreeRoot.EstimatedRows * nodeDegrees;
            var estimatedNodeUnitSize = nodeUnitCandidate.TreeRoot.EstimatedRows *
                                        nodeUnitCandidate.SqlEstimatedEdgeDegrees;
            var componentSize     = Cardinality;
            var estimatedCompSize = SqlEstimatedSize;

            var node = nodeUnitCandidate.TreeRoot;

            // If the node is already in the component, then only multiply the degree to get the size
            double nodeUnitActualSize;
            double newCompEstSize;

            if (MaterializedNodeSplitCount[node] > 0)
            {
                nodeUnitActualSize = nodeDegrees;
                var cEstEdge = Math.Pow(1000, EdgeMaterilizedDict.Count(e => !e.Value));
                var cSize    = SqlEstimatedSize / cEstEdge;
                var nSize    = node.EstimatedRows;
                if (nSize > cSize)
                {
                    newCompEstSize = estimatedNodeUnitSize * cEstEdge * estimatedSelectivity;
                }
                else
                {
                    newCompEstSize = SqlEstimatedSize * Math.Pow(1000, nodeUnitCandidate.UnmaterializedEdges.Count) * estimatedSelectivity;
                }
            }
            else
            {
                nodeUnitActualSize = nodeUnitSize;
                newCompEstSize     = SqlEstimatedSize * estimatedNodeUnitSize * estimatedSelectivity;
            }
            newCompEstSize = newCompEstSize < 1.0 ? 1.0 : newCompEstSize;



            bool firstJoin = MaterializedNodeSplitCount.Count == 2 &&
                             MaterializedNodeSplitCount.All(e => e.Value == 0);

            // Update TableRef
            double loopJoinOuterThreshold = 1e4; //1e6;
            double sizeFactor             = 5;   //1000;
            double maxMemory = 1e8;
            double loopCost  = componentSize * Math.Log(nodeUnitCandidate.TreeRoot.EstimatedRows, 512) * 0.20;
            double hashCost  = componentSize + nodeUnitSize;
            double cost;

            // Loop Join
            if (
                nodeUnitCandidate.MaterializedEdges.Count == 0 && // the joins are purely leaf to sink join
                (
                    //componentSize < loopJoinOuterThreshold ||     // the outer table is relatively small
                    loopCost < hashCost ||
                    (DeltaMemory + componentSize > maxMemory && DeltaMemory + nodeUnitSize > maxMemory) // memory is in pressure
                )
                )
            {
                if (firstJoin)
                {
                    RightestTableRefSize = nodeUnitCandidate.TreeRoot.EstimatedRows;
                    RightestTableAlias   = GetNodeRefName(node);
                }
                TotalMemory             = DeltaMemory;
                SqlEstimatedTotalMemory = SqlEstimatedDeltaMemory;
                //joinTable.JoinHint = JoinHint.Loop;
                SqlEstimatedSize = estimatedCompSize * estimatedNodeUnitSize /
                                   nodeUnitCandidate.TreeRoot.TableRowCount;

                cost     = loopCost; //componentSize*Math.Log(nodeUnitCandidate.TreeRoot.EstimatedRows, 512);
                TableRef = new WParenthesisTableReference
                {
                    Table = new WQualifiedJoin
                    {
                        FirstTableRef  = TableRef,
                        SecondTableRef =
                            nodeUnitCandidate.ToTableReference(GetNodeRefName(nodeUnitCandidate.TreeRoot), metaData),
                        JoinCondition     = joinCondition,
                        QualifiedJoinType = QualifiedJoinType.Inner,
                        JoinHint          = JoinHint.Loop
                    }
                };
            }
            // Hash Join
            else
            {
                cost = hashCost;//componentSize + nodeUnitSize;
                WBooleanExpression adjustedJoincondition;
                double             adjustedSqlEstimatedSize;
                WTableReference    buildTableReference;
                WTableReference    probeTableReference;
                if (firstJoin)
                {
                    var nodeInComp = MaterializedNodeSplitCount.Keys.First(e => e != node);
                    if (nodeUnitSize < componentSize)
                    {
                        buildTableReference = AdjustEstimation(nodeUnitCandidate, GetNodeRefName(node), metaData, out adjustedJoincondition,
                                                               out adjustedSqlEstimatedSize);
                        probeTableReference     = TableRef;
                        TotalMemory             = DeltaMemory = nodeUnitSize;
                        SqlEstimatedTotalMemory = SqlEstimatedDeltaMemory = estimatedNodeUnitSize;
                        RightestTableRefSize    = nodeInComp.EstimatedRows;
                        RightestTableAlias      = GetNodeRefName(nodeInComp);
                    }
                    else
                    {
                        RightestTableRefSize = nodeInComp.EstimatedRows;
                        RightestTableAlias   = GetNodeRefName(nodeInComp);
                        buildTableReference  = AdjustEstimation(this, out adjustedJoincondition, out adjustedSqlEstimatedSize);
                        probeTableReference  =
                            nodeUnitCandidate.ToTableReference(GetNodeRefName(nodeUnitCandidate.TreeRoot), metaData);
                        TotalMemory             = DeltaMemory = componentSize;
                        SqlEstimatedTotalMemory = SqlEstimatedDeltaMemory = SqlEstimatedSize;
                        RightestTableRefSize    = nodeUnitCandidate.TreeRoot.EstimatedRows;
                        RightestTableAlias      = GetNodeRefName(node);
                    }
                }
                // Left Deep
                else if (componentSize * sizeFactor < nodeUnitSize)
                {
                    // Adjust estimation in sql server
                    buildTableReference = AdjustEstimation(this, out adjustedJoincondition, out adjustedSqlEstimatedSize);
                    probeTableReference =
                        nodeUnitCandidate.ToTableReference(GetNodeRefName(nodeUnitCandidate.TreeRoot), metaData);
                    var curDeltaMemory = componentSize;
                    TotalMemory = DeltaMemory + curDeltaMemory;
                    DeltaMemory = curDeltaMemory;
                    var curDeltaEstimateMemory = SqlEstimatedSize;
                    SqlEstimatedTotalMemory = SqlEstimatedDeltaMemory + curDeltaEstimateMemory;
                    SqlEstimatedDeltaMemory = curDeltaEstimateMemory;


                    RightestTableAlias   = GetNodeRefName(node);
                    RightestTableRefSize = nodeUnitCandidate.TreeRoot.EstimatedRows;
                }
                // Right Deep
                else
                {
                    buildTableReference = AdjustEstimation(nodeUnitCandidate, GetNodeRefName(node), metaData, out adjustedJoincondition,
                                                           out adjustedSqlEstimatedSize);
                    probeTableReference = TableRef;

                    TotalMemory             += nodeUnitSize;
                    DeltaMemory              = TotalMemory;
                    SqlEstimatedTotalMemory += estimatedNodeUnitSize;
                    SqlEstimatedDeltaMemory  = SqlEstimatedTotalMemory;
                }
                newCompEstSize *= adjustedSqlEstimatedSize;
                TableRef        = new WParenthesisTableReference
                {
                    Table =
                        new WQualifiedJoin
                    {
                        FirstTableRef     = buildTableReference,
                        SecondTableRef    = probeTableReference,
                        JoinCondition     = WBooleanBinaryExpression.Conjunction(joinCondition, adjustedJoincondition),
                        QualifiedJoinType = QualifiedJoinType.Inner,
                        JoinHint          = JoinHint.Hash
                    }
                };
                SqlEstimatedSize = newCompEstSize < 1.0 ? 1.0 : newCompEstSize;
            }

            //Update Size
            Cardinality *= nodeUnitActualSize * joinSelectivity;

            // Debug
#if DEBUG
            //foreach (var item in MaterializedNodeSplitCount.Where(e => e.Key != node))
            //{
            //    Trace.Write(item.Key.RefAlias + ",");
            //}
            //Trace.Write(node.RefAlias);
            //Trace.Write(" Size:" + Cardinality + " Cost:" + cost);
            //Trace.Write(" Method:" + ((TableRef as WParenthesisTableReference).Table as WQualifiedJoin).JoinHint);
            //Trace.WriteLine(" --> Total Cost:" + Cost);
#endif


            // Update Cost
            Cost += cost;
        }