Skip to content

bizo/spark-plug

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

63 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

spark-plug

Build Status

A scala driver for launching Amazon EMR jobs

why?

We run a lot of reports. In the past, these have been kicked off by bash scripts that typically do things like date math, copy scripts and config files to s3 before calling to the amazon elastic-mapreduce command line client to launch the job. The emr client invocation ends up being dozen of lines of bash code adding each step and passing arguments.

It's been a pain to share defaults or add any abstraction over common job steps. Additionally, performing date arithmetic and conditionally adding EMR steps can be a pain. Lastly, the EMR client offers less control over certain options available from the EMR API.

simple example

val flow = JobFlow(
  name      = s"${stage}: analytics report [${date}]",
  cluster   = Master() + Core(8) + Spot(10),
  bootstrap = Seq(MemoryIntensive),
  steps     = Seq(
    SetupDebugging(),
    new HiveStep("s3://bucket/location/report.sql",
      Map("YEAR" -> year, "MONTH" -> month, "DAY" -> day))
  )
)

val id = Emr.run(flow)(ClusterDefaults(hadoop="1.0.3"))
println(id)

API documentation

download

Available in Maven Central as com.bizo spark-plug_2.10

About

scala driver for launching Amazon EMR jobs

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Scala 100.0%