1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
|
---
layout: global
title: Cluster Mode Overview
---
This document gives a short overview of how Spark runs on clusters, to make it easier to understand
the components involved. Read through the [application submission guide](submitting-applications.html)
to learn about launching applications on a cluster.
# Components
Spark applications run as independent sets of processes on a cluster, coordinated by the `SparkContext`
object in your main program (called the _driver program_).
Specifically, to run on a cluster, the SparkContext can connect to several types of _cluster managers_
(either Spark's own standalone cluster manager, Mesos or YARN), which allocate resources across
applications. Once connected, Spark acquires *executors* on nodes in the cluster, which are
processes that run computations and store data for your application.
Next, it sends your application code (defined by JAR or Python files passed to SparkContext) to
the executors. Finally, SparkContext sends *tasks* to the executors to run.
<p style="text-align: center;">
<img src="img/cluster-overview.png" title="Spark cluster components" alt="Spark cluster components" />
</p>
There are several useful things to note about this architecture:
1. Each application gets its own executor processes, which stay up for the duration of the whole
application and run tasks in multiple threads. This has the benefit of isolating applications
from each other, on both the scheduling side (each driver schedules its own tasks) and executor
side (tasks from different applications run in different JVMs). However, it also means that
data cannot be shared across different Spark applications (instances of SparkContext) without
writing it to an external storage system.
2. Spark is agnostic to the underlying cluster manager. As long as it can acquire executor
processes, and these communicate with each other, it is relatively easy to run it even on a
cluster manager that also supports other applications (e.g. Mesos/YARN).
3. The driver program must listen for and accept incoming connections from its executors throughout
its lifetime (e.g., see [spark.driver.port in the network config
section](configuration.html#networking)). As such, the driver program must be network
addressable from the worker nodes.
4. Because the driver schedules tasks on the cluster, it should be run close to the worker
nodes, preferably on the same local area network. If you'd like to send requests to the
cluster remotely, it's better to open an RPC to the driver and have it submit operations
from nearby than to run a driver far away from the worker nodes.
# Cluster Manager Types
The system currently supports three cluster managers:
* [Standalone](spark-standalone.html) -- a simple cluster manager included with Spark that makes it
easy to set up a cluster.
* [Apache Mesos](running-on-mesos.html) -- a general cluster manager that can also run Hadoop MapReduce
and service applications.
* [Hadoop YARN](running-on-yarn.html) -- the resource manager in Hadoop 2.
* [Kubernetes (experimental)](https://github.com/apache-spark-on-k8s/spark) -- In addition to the above,
there is experimental support for Kubernetes. Kubernetes is an open-source platform
for providing container-centric infrastructure. Kubernetes support is being actively
developed in an [apache-spark-on-k8s](https://github.com/apache-spark-on-k8s/) Github organization.
For documentation, refer to that project's README.
# Submitting Applications
Applications can be submitted to a cluster of any type using the `spark-submit` script.
The [application submission guide](submitting-applications.html) describes how to do this.
# Monitoring
Each driver program has a web UI, typically on port 4040, that displays information about running
tasks, executors, and storage usage. Simply go to `http://<driver-node>:4040` in a web browser to
access this UI. The [monitoring guide](monitoring.html) also describes other monitoring options.
# Job Scheduling
Spark gives control over resource allocation both _across_ applications (at the level of the cluster
manager) and _within_ applications (if multiple computations are happening on the same SparkContext).
The [job scheduling overview](job-scheduling.html) describes this in more detail.
# Glossary
The following table summarizes terms you'll see used to refer to cluster concepts:
<table class="table">
<thead>
<tr><th style="width: 130px;">Term</th><th>Meaning</th></tr>
</thead>
<tbody>
<tr>
<td>Application</td>
<td>User program built on Spark. Consists of a <em>driver program</em> and <em>executors</em> on the cluster.</td>
</tr>
<tr>
<td>Application jar</td>
<td>
A jar containing the user's Spark application. In some cases users will want to create
an "uber jar" containing their application along with its dependencies. The user's jar
should never include Hadoop or Spark libraries, however, these will be added at runtime.
</td>
</tr>
<tr>
<td>Driver program</td>
<td>The process running the main() function of the application and creating the SparkContext</td>
</tr>
<tr>
<td>Cluster manager</td>
<td>An external service for acquiring resources on the cluster (e.g. standalone manager, Mesos, YARN)</td>
</tr>
<tr>
<td>Deploy mode</td>
<td>Distinguishes where the driver process runs. In "cluster" mode, the framework launches
the driver inside of the cluster. In "client" mode, the submitter launches the driver
outside of the cluster.</td>
</tr>
<tr>
<td>Worker node</td>
<td>Any node that can run application code in the cluster</td>
</tr>
<tr>
<td>Executor</td>
<td>A process launched for an application on a worker node, that runs tasks and keeps data in memory
or disk storage across them. Each application has its own executors.</td>
</tr>
<tr>
<td>Task</td>
<td>A unit of work that will be sent to one executor</td>
</tr>
<tr>
<td>Job</td>
<td>A parallel computation consisting of multiple tasks that gets spawned in response to a Spark action
(e.g. <code>save</code>, <code>collect</code>); you'll see this term used in the driver's logs.</td>
</tr>
<tr>
<td>Stage</td>
<td>Each job gets divided into smaller sets of tasks called <em>stages</em> that depend on each other
(similar to the map and reduce stages in MapReduce); you'll see this term used in the driver's logs.</td>
</tr>
</tbody>
</table>
|
