public void BuildGraphFromQuery(DryadLINQApp app, Query query)
{
// set configurable properties
int highThreshold = app.GetMaxAggregateInputs();
int lowThreshold = 16;
UInt64 highDataThreshold = (UInt64)app.GetAggregateThreshold();
UInt64 lowDataThreshold = (3 * highDataThreshold) / 4;
UInt64 maxSingleDataThreshold = highDataThreshold / 2;
int aggFilterThreshold = app.GetMaxAggregateFilterInputs();
// use a graph stage map to store the vertices as they are created, grouped by stage.
Dictionary <int, GraphStageInfo> graphStageMap = new Dictionary <int, GraphStageInfo>();
DryadLogger.LogInformation("Building graph");
//
// Create a set of vertices for each vertex (stage) in the query plan
//
DryadLogger.LogInformation("Adding vertices");
foreach (KeyValuePair <int, Vertex> kvp in query.queryPlan)
{
Vertex v = kvp.Value;
GraphStageInfo value = null;
if (!graphStageMap.TryGetValue(v.uniqueId, out value))
{
DryadLogger.LogInformation(String.Format("Adding vertices for stage {0}", v.name));
CreateVertexSet(v, app, query, graphStageMap);
}
}
//
// Add dynamic stage managers
//
DryadLogger.LogInformation("Build Graph From Query", "Adding stage managers");
foreach (KeyValuePair <int, GraphStageInfo> kvp in graphStageMap)
{
Vertex v = kvp.Value.vertex;
//
//There are no dynamic managers
//
if (v.dynamicManager == null)
{
continue;
}
DrStageManager newManager = kvp.Value.stageManager; // newManager
DrGraphParameters parameters = app.GetGraph().GetParameters();
string stdVertexName = "MW";
string cpyVertexName = "CP";
if (v.type != Vertex.Type.INPUTTABLE && v.type != Vertex.Type.CONCAT)
{
if (v.dynamicManager.type == DynamicManager.Type.SPLITTER)
{
if (v.info.predecessors.Length == 1)
{
DrPipelineSplitManager splitter = new DrPipelineSplitManager();
newManager.AddDynamicConnectionManager(graphStageMap[v.info.predecessors[0].uniqueId].stageManager, splitter);
}
else
{
DrSemiPipelineSplitManager splitter = new DrSemiPipelineSplitManager();
newManager.AddDynamicConnectionManager(graphStageMap[v.info.predecessors[0].uniqueId].stageManager, splitter);
}
}
else if (v.dynamicManager.type == DynamicManager.Type.PARTIALAGGR)
{
DrDynamicAggregateManager dynamicMerge = new DrDynamicAggregateManager();
dynamicMerge.SetGroupingSettings(0, 0);
dynamicMerge.SetMachineGroupingSettings(2, aggFilterThreshold);
dynamicMerge.SetDataGroupingSettings(lowDataThreshold, highDataThreshold, maxSingleDataThreshold);
dynamicMerge.SetSplitAfterGrouping(true);
foreach (Predecessor p in v.info.predecessors)
{
newManager.AddDynamicConnectionManager(graphStageMap[p.uniqueId].stageManager, dynamicMerge);
}
}
else if (v.dynamicManager.type == DynamicManager.Type.FULLAGGR ||
v.dynamicManager.type == DynamicManager.Type.HASHDISTRIBUTOR)
{
int idx = 0;
int sz = v.dynamicManager.assemblyNames == null ? 0 : v.dynamicManager.assemblyNames.Length;
DrDynamicAggregateManager dynamicMerge = new DrDynamicAggregateManager();
if (v.dynamicManager.type == DynamicManager.Type.FULLAGGR || sz > 1)
{
dynamicMerge = new DrDynamicAggregateManager();
string name = v.dynamicManager.methodNames[idx];
DrManagerBase newStage = new DrManagerBase(app.GetGraph(), name);
DrActiveVertex mergeVertex = new DrActiveVertex(newStage, parameters.m_defaultProcessTemplate, parameters.m_defaultVertexTemplate);
mergeVertex.AddArgument(stdVertexName);
mergeVertex.AddArgument(v.dynamicManager.assemblyNames[idx]);
mergeVertex.AddArgument(v.dynamicManager.classNames[idx]);
mergeVertex.AddArgument(v.dynamicManager.methodNames[idx]);
idx++;
dynamicMerge.SetInternalVertex(mergeVertex);
dynamicMerge.SetGroupingSettings(0, 0);
dynamicMerge.SetPodGroupingSettings(lowThreshold, highThreshold);
dynamicMerge.SetDataGroupingSettings(lowDataThreshold,
highDataThreshold,
maxSingleDataThreshold);
dynamicMerge.SetMaxAggregationLevel(v.dynamicManager.aggregationLevels);
}
if (v.dynamicManager.type == DynamicManager.Type.FULLAGGR)
{
newManager.AddDynamicConnectionManager(graphStageMap[v.info.predecessors[0].uniqueId].stageManager, dynamicMerge);
}
else
{
string name = v.dynamicManager.methodNames[idx];
DrManagerBase newStage = new DrManagerBase(app.GetGraph(), name);
DrActiveVertex distributeVertex = new DrActiveVertex(newStage, parameters.m_defaultProcessTemplate, parameters.m_defaultVertexTemplate);
distributeVertex.AddArgument(stdVertexName);
distributeVertex.AddArgument(v.dynamicManager.assemblyNames[idx]);
distributeVertex.AddArgument(v.dynamicManager.classNames[idx]);
distributeVertex.AddArgument(v.dynamicManager.methodNames[idx]);
idx++;
DrDynamicDistributionManager dynamicHashDistribute =
new DrDynamicDistributionManager(distributeVertex, dynamicMerge);
dynamicHashDistribute.SetDataPerVertex(highDataThreshold * 2); // 2GB
newManager.AddDynamicConnectionManager(graphStageMap[v.info.predecessors[0].uniqueId].stageManager, dynamicHashDistribute);
}
}
else if (v.dynamicManager.type == DynamicManager.Type.RANGEDISTRIBUTOR)
{
DrStageManager splitManager = graphStageMap[v.dynamicManager.splitVertexId].stageManager;
DrDynamicRangeDistributionManager drdm = new DrDynamicRangeDistributionManager(splitManager, v.dynamicManager.sampleRate);
drdm.SetDataPerVertex(highDataThreshold * 2); // 2GB
newManager.AddDynamicConnectionManager(graphStageMap[v.info.predecessors[0].uniqueId].stageManager, drdm);
}
else if (v.dynamicManager.type == DynamicManager.Type.BROADCAST)
{
// the copy vertex
int bcastNumber = 0;
string nameString = String.Format("CP__{0}", bcastNumber++);
DrManagerBase newStage = new DrManagerBase(app.GetGraph(), nameString);
DrActiveVertex copyVertex =
new DrActiveVertex(newStage,
parameters.m_defaultProcessTemplate,
parameters.m_defaultVertexTemplate);
copyVertex.AddArgument(cpyVertexName);
DrDynamicBroadcastManager bcast = new DrDynamicBroadcastManager(copyVertex);
newManager.AddDynamicConnectionManager(graphStageMap[v.info.predecessors[0].uniqueId].stageManager, bcast);
}
else if (v.dynamicManager.type != DynamicManager.Type.NONE)
{
DryadLogger.LogWarning(String.Format("Dynamic manager type {0} not supported yet", v.dynamicManager.type));
}
}
}
//
// Add all the edges
//
DryadLogger.LogInformation("Build Graph From Query", "Adding edges");
foreach (KeyValuePair <int, GraphStageInfo> kvp in graphStageMap)
{
AddEdges(kvp.Value, graphStageMap);
}
//
// Register the actual created vertices with the graph
//
MaterializeToManagers(graphStageMap);
}
private void CreateVertexSet(Vertex v, DryadLINQApp app, Query query, Dictionary <int, GraphStageInfo> graphStageMap)
{
SortedDictionary <int, Vertex> queryPlan = query.queryPlan;
DrVertexSet nodes = null;
DrStageManager newManager = null;
app.GetGraph().GetParameters();
DrGraphParameters parameters = app.GetGraph().GetParameters();
string stdVertexName = "MW";
if (v.type == Vertex.Type.INPUTTABLE)
{
DrInputStreamManager input = CreateInputNode(app, v.info, v.name);
newManager = input.GetStageManager();
nodes = CreateVertexSet(app.GetGraph(), input);
}
else if (v.type == Vertex.Type.OUTPUTTABLE)
{
DrOutputStreamManager output = CreateOutputNode(app, v.info, v.name);
newManager = output.GetStageManager();
nodes = CreateVertexSet(app.GetGraph(), output, v.partitions);
}
else if (v.type == Vertex.Type.CONCAT)
{
newManager = new DrManagerBase(app.GetGraph(), v.name);
// the set of nodes in a concat is just the set of nodes in the predecessor stages concatenated
nodes = new DrVertexSet();
foreach (Predecessor p in v.info.predecessors)
{
GraphStageInfo value = null;
if (graphStageMap.TryGetValue(p.uniqueId, out value))
{
nodes.InsertRange(nodes.Count, value.members);
}
else
{
throw new LinqToDryadException(String.Format("Concat: Failed to find predecessor {0} in graph stage map", p.uniqueId));
}
}
}
else
{
newManager = new DrManagerBase(app.GetGraph(), v.name);
DrVertex vertex;
if (v.type == Vertex.Type.TEE)
{
DrTeeVertex teeVertex = new DrTeeVertex(newManager);
vertex = teeVertex;
}
else
{
DrActiveVertex activeVertex = new DrActiveVertex(newManager, parameters.m_defaultProcessTemplate, parameters.m_defaultVertexTemplate);
activeVertex.AddArgument(stdVertexName);
activeVertex.AddArgument(v.info.assemblyName);
activeVertex.AddArgument(v.info.className);
activeVertex.AddArgument(v.info.methodName);
vertex = activeVertex;
}
nodes = CreateVertexSet(app.GetGraph(), vertex, v.partitions);
if (v.machines != null && v.machines.Length != 0 && v.type != Vertex.Type.TEE)
{
for (int i = 0; i < v.partitions; i++)
{
DrResource r = app.GetUniverse().LookUpResource(v.machines[i]);
if (r != null)
{
DrVertex baseVertex = nodes[i];
DrActiveVertex activeVertex = baseVertex as DrActiveVertex;
activeVertex.GetAffinity().SetHardConstraint(true);
activeVertex.GetAffinity().AddLocality(r);
}
}
}
FileStream mapfile = app.GetIdentityMapFile();
if (mapfile != null)
{
for (int i = 0; i < v.partitions; i++)
{
DrVertex jmv = nodes[i];
string message = String.Format("Mapping {0}[{1}] to {2}\n", v.uniqueId, i, jmv.GetId());
System.Text.ASCIIEncoding encoding = new System.Text.ASCIIEncoding();
mapfile.Seek(0, SeekOrigin.End);
mapfile.Write(encoding.GetBytes(message), 0, message.Length);
}
}
}
graphStageMap.Add(v.uniqueId, new GraphStageInfo(nodes, v, newManager));
}