Databases Reference
In-Depth Information
CHAPTER 7
Using Non-JVM Languages with Storm
Sometimes you want to use languages that aren't based on the JVM to implement a
Storm project, either because you feel more comfortable with another language or you
want to use a library written in that language.
Storm is implemented in Java, and all the spouts and bolts that you've seen in this topic
were written in Java as well. So is it possible to use languages like Python, Ruby, or
even JavaScript to write spouts and bolts? The answer is yes! It is possible using some-
thing called the
multilang protocol
.
The multilang protocol is a special protocol implemented in Storm that uses
standard
input
and
standard output
as a channel of communication with a process that does the
job of a spout or a bolt. Messages are passed through this channel encoded as JSON
or as lines of plain text.
Let's take a look at a simple example of a spout and a bolt in a non-JVM language.
You'll have a spout that generates numbers from 1 to 10,000 and a bolt that filters for
prime numbers, both written in PHP.
In this example, we check for prime numbers in a naive way. There are
much better implementations, but they are also more complex and out
of the scope of this example.
There is an official implementation of a PHP DSL for Storm. In this chapter, we'll show
our implementation as an example. First of all, define the topology.
...
TopologyBuilder
builder
=
new
TopologyBuilder
();
builder
.
setSpout
(
"numbers-generator"
,
new
NumberGeneratorSpout
(
1
,
10000
));
builder
.
setBolt
(
"prime-numbers-filter"
,
new
PrimeNumbersFilterBolt
()).
shuffleGrouping
(
"numbers-generator"
);
StormTopology
topology
=
builder
.
createTopology
();
...
