graphx-programming-guide.md

layout: global
title: GraphX Programming Guide

• This will become a table of contents (this text will be scraped). {:toc}

Overview

GraphX is the new (alpha) Spark API for graphs and graph-parallel computation. At a high-level, GraphX extends the Spark RDD by introducing the Resilient Distributed property Graph (RDG): a directed graph with properties attached to each vertex and edge. To support graph computation, GraphX exposes a set of functions (e.g., mapReduceTriplets) as well as an optimized variant of the Pregel API. In addition, GraphX includes a growing collection of graph algorithms and builders to simplify graph analytics tasks.

Background on Graph-Parallel Computation

From social networks to language modeling, the growing scale and importance of graph data has driven the development of numerous new graph-parallel systems (e.g., Giraph and GraphLab). By restricting the types of computation that can be expressed and introducing new techniques to partition and distribute graphs, these systems can efficiently execute sophisticated graph algorithms orders of magnitude faster than more general data-parallel systems.

However, the same restrictions that enable these substantial performance gains also make it difficult to express many of the important stages in a typical graph-analytics pipeline: constructing the graph, modifying its structure, or expressing computation that spans multiple graphs. As a consequence, existing graph analytics pipelines compose graph-parallel and data-parallel systems, leading to extensive data movement and duplication and a complicated programming model.

The goal of the GraphX project is to unify graph-parallel and data-parallel computation in one system with a single composable API. The GraphX API enables users to view data both as a graph and as collection (i.e., RDDs) without data movement or duplication. By incorporating recent advances in graph-parallel systems, GraphX is able to optimize the execution of graph operations.

GraphX Replaces the Spark Bagel API

Prior to the release of GraphX, graph computation in Spark was expressed using Bagel, an implementation of the Pregel API. GraphX improves upon Bagel by exposing a richer property graph API, a more streamlined version of the Pregel abstraction, and system optimizations to improve performance and reduce memory overhead. While we plan to eventually deprecate the Bagel, we will continue to support the Bagel API and Bagel programming guide. However, we encourage Bagel users to explore the new GraphX API and comment on issues that may complicate the transition from Bagel.

Getting Started

To get started you first need to import Spark and GraphX into your project. This can be done by importing the following:

{% highlight scala %} import org.apache.spark._ import org.apache.spark.graphx._ {% endhighlight %}

If you are not using the Spark shell you will also need a Spark context.

The Property Graph

The property graph is a directed multigraph graph with user defined objects attached to each vertex and edge. As a multigraph it is possible for multiple edges to have the same source and destination vertex. This can be useful when there are multiple relationships between the same vertices. Like RDDs, property graphs are immutable, distributed, and fault-tolerant. Vertices are keyed by their vertex identifier (VertexId) which is a unique 64-bit long. Similarly, edges have corresponding source and destination vertex identifiers. Unlike other systems, GraphX does not impose any ordering or constraints on the vertex identifiers.

The property graph is parameterized over the vertex VD and edge ED types. These are the types of the objects associated with each vertex and edge respectively. In some cases it can be desirable to have vertices of different types. However, this can be accomplished through inheritance.

GraphX optimizes the representation of VD and ED when they are plain old data-types (e.g., int, double, etc...) reducing the in memory footprint.

Logically the property graph corresponds to a pair of typed collections (RDDs) encoding the properties for each vertex and edge:

{% highlight scala %} class Graph[VD: ClassTag, ED: ClassTag] { val vertices: RDD[(VertexId, VD)] val edges: RDD[Edge[ED]] // ... } {% endhighlight %}

Note that the vertices and edges of the graph are actually of type VertexRDD[VD] and EdgeRDD[ED] respectively. These types extend and are optimized versions of RDD[(VertexId, VD)] and RDD[Edge[ED]].

For example, we might construct a property graph consisting of various collaborators on the GraphX project. The vertex property contains the username and occupation and the edge property contains a string describing the relationships between the users.

The resulting graph would have the type signature:

{% highlight scala %} val userGraph: Graph[(String, String), String] {% endhighlight %}

There are numerous ways to construct a property graph from raw files, RDDs, and even synthetic generators and these are discussed in more detail in the section on graph builders. Probably the most general method is to use the graph singleton. For example the following code constructs a graph from a collection of RDDs:

{% highlight scala %} // Assume the SparkContext has already been constructed val sc: SparkContext // Create an RDD for the vertices val users: RDD[(VertexId, (String, String))] = sc.parallelize(Array((3, ("rxin", "student")), (7, ("jgonzal", "postdoc")), (5, ("franklin", "prof")), (2, ("istoica", "prof")))) // Create an RDD for edges val relationships: RDD[Edge[String]] = sc.parallelize(Array(Edge(3, 7, "collab"), Edge(5, 3, "advisor"), Edge(2, 5, "colleague"), Edge(5, 7, "pi")) // Define a default user in case there are relationship with missing user val defaultUser = ("John Doe", "Missing") // Build the initial Graph val graph = Graph(users, relationships, defaultUser) {% endhighlight %}

In the above example we make use of the Edge case class. Edges have a srcId and a dstId corresponding to the source and destination vertex identifiers. In addition, the Edge class contains the attr member which contains the edge property.

We can deconstruct a graph into the respective vertex and edge views by using the graph.vertices and graph.edges members respectively.

{% highlight scala %} val graph: Graph[(String, String), String] // Constructed from above // Count all users which are postdocs graph.vertices.filter { case (id, (name, pos)) => pos == "postdoc"}.count // Count all the edges where src > dst graph.edges.filter(e => e.srcId > e.dstId).count {% endhighlight %}

Note that graph.vertices returns an RDD[(VertexId, (String, String))] and so we must use the scala case expression to deconstruct the tuple. Alternatively, graph.edges returns an RDD containing Edge[String] objects. We could have also used the case class type constructor as in the following: {% highlight scala %} graph.edges.filter { case Edge(src, dst, prop) => src < dst }.count {% endhighlight %}

In addition to the vertex and edge views of the property graph, GraphX also exposes a triplet view. The triplet view logically joins the vertex and edge properties yielding an RDD[EdgeTriplet[VD, ED]] consisting of EdgeTriplet. This join can be expressed in the following SQL expression:

{% highlight sql %} SELECT src.id, dst.id, src.attr, e.attr, dst.attr FROM edges AS e LEFT JOIN vertices AS src, vertices AS dst ON e.srcId = src.Id AND e.dstId = dst.Id {% endhighlight %}

or graphically as:

The EdgeTriplet class extends the Edge class by adding the srcAttr and dstAttr members which contain the source and destination properties respectively. We can use the triplet view of a graph to render a collection of strings describing relationships between users.

{% highlight scala %} val graph: Graph[(String, String), String] // Constructed from above // Use the triplets view to create an RDD of facts. val facts: RDD[String] = graph.triplets.map(et => et.srcAttr._1 + " is the " + et.attr + " of " et.dstAttr) {% endhighlight %}

Graph Operators

Just as RDDs have basic operations like map, filter, and reduceByKey, property graphs also have a collection of basic operators that take user defined function and produce new graphs with transformed properties and structure. The core operators that have optimized implementations are defined in Graph.scala and convenient operators that are expressed as a compositions of the core operators are defined in 'GraphOps.scala'. However, thanks to Scala implicits the operators in GraphOps.scala are automatically available as members of Graph.scala. For example, we can compute the in-degree of each vertex (defined in 'GraphOps.scala') by the following:

{% highlight scala %} val graph: Graph[(String, String), String] // Use the implicit GraphOps.inDegrees operator val indDegrees: VertexRDD[Int] = graph.inDegrees {% endhighlight %}

The reason for differentiating between core graph operations and GraphOps is to be able to support various graph representations in the future.

Property Operators

In direct analogy to the RDD map operator, the property graph contains the following:

{% highlight scala %} def mapVertices[VD2](map:%20(VertexID,%20VD) => VD2): Graph[VD2, ED] def mapEdgesED2: Graph[VD, ED2] def mapTripletsED2: Graph[VD, ED2] {% endhighlight %}

Each of these operators yields a new graph with the vertex or edge properties modified by the user defined map function.

Note that in all cases the graph structure is unaffected. This is a key feature of these operators which allows the resulting graph to reuse the structural indicies and the unaffected properties of the original graph. While graph.mapVertices(mapUDF) is logically equivalent to the following, the following does not preserve the structural indicies and would not benefit from the substantial system optimizations in GraphX. {% highlight scala %} val newVertices = graph.vertices.map { case (id, attr) => (id, mapUdf(id, attr))} val newGraph = Graph(newVertices, graph.edges) {% endhighlight %}

These operators are often used to initialize the graph for a particular computation or project away unnecessary properties. For example, given a graph with the out-degrees as the vertex properties (we describe how to construct such a graph later) we initialize for PageRank:

{% highlight scala %} // Given a graph where the vertex property is the out-degree val inputGraph: Graph[Int, String] // Construct a graph where each edge contains the weight // and each vertex is the initial PageRank val outputGraph: Graph[Double, Double] = inputGraph.mapTriplets(et => 1.0/et.srcAttr).mapVertices(v => 1.0) {% endhighlight %}

Structural Operators

Currently GraphX supports only a simple set of commonly used structural operators and we expect to add more in the future. The following is a list of the basic structural operators.

{% highlight scala %} def reverse: Graph[VD, ED]

def subgraph(epred: EdgeTriplet[VD,ED] => Boolean = (x => true), vpred: (VertexID, VD) => Boolean = ((v,d) => true) ): Graph[VD, ED]

def maskVD2, ED2: Graph[VD, ED]

def groupEdges(merge: (ED, ED) => ED): Graph[VD,ED] {% endhighlight %}

The reverse operator returns a new graph with all the edge directions reversed. This can be useful when, for example, trying to compute the inverse PageRank. Because the reverse operation does not modify vertex or edge properties or change the number of edges, it can be implemented efficiently without data-movement or duplication.

The subgraph operator takes vertex and edge predicates and returns the graph containing only the vertices that satisfy the vertex predicate (evaluate to true) and edges that satisfy the edge predicate and connect vertices that satisfy the vertex predicate. The subgraph operator can be used in number of situations to restrict the graph to the vertices and edges of interest or eliminate broken links. For example in the following code we remove broken links:

{% highlight scala %} val users: RDD[(VertexId, (String, String))] val edges: RDD[Edge[String]] // Define a default user in case there are relationship with missing user val defaultUser = ("John Doe", "Missing") // Build the initial Graph val graph = Graph(users, relationships, defaultUser) // Remove missing vertices as well as the edges to connected to them val validGraph = graph.subgraph((id, attr) => attr._2 != "Missing") {% endhighlight %}

The mask operators returns the subgraph containing only the vertices and edges that are found in the input graph. This can be used in conjunction with the subgraph operator to restrict a graph based on the properties in another related graph. For example, we might run connected components using the graph with missing vertices and then restrict the answer to the valid subgraph.

{% highlight scala %} // Run Connected Components val ccGraph = graph.connectedComponents() // Remove missing vertices as well as the edges to connected to them val validGraph = graph.subgraph((id, attr) => attr._2 != "Missing") // Restrict the answer to the valid subgraph val validCCGraph = ccGraph.mask(validGraph) {% endhighlight %}

The groupEdges operator merges parallel edges: duplicate edges between pairs of vertices. In many numerical applications parallel edges can be added (their weights combined) into a single edge thereby reducing the graph size in memory as well as the cost of computation.

Join Operators

The ability to move between graph and collection views of data is a key part of GraphX. In many cases it is necessary to bring data from external collections into the graph. For example, we might have extra user properties that we want to merge with an existing graph or we might want to pull vertex properties from one graph into another. These tasks can be accomplished using the join operators. Below we list the key join operators:

{% highlight scala %} def joinVertices[U](table:%20RDD[(VertexID,%20U)])(mapFunc: (VertexID, VD, U) => VD)

Graph[VD, ED] def outerJoinVertices[U, VD2](table:%20RDD[(VertexID,%20U)])(mapFunc: (VertexID, VD, Option[U]) => VD2)

Graph[VD2, ED] {% endhighlight %}

Map Reduce Triplets (mapReduceTriplets)