What is configuartion required to get data from object storage by SWIFT in Spark

772 views Asked by At

I go through document but still it is very much confusing how to get data from swift.

I configured swift in my one linux machine. By using below command I am able to get container list,

swift -A https://acc.objectstorage.softlayer.net/auth/v1.0/ -U username -K passwordkey list

I seen many blog for blumix(https://console.ng.bluemix.net/docs/services/AnalyticsforApacheSpark/index-gentopic1.html#genTopProcId2) and written the below code

sc.textFile("swift://container.myacct/file.xml")

I am looking to integrate in java spark. Where need to configure object storage credential in java code. Is there any sample code or blog?

1

There are 1 answers

0
Sanjay.Joshi On

This notebook illustrates a number of ways to load data using the Scala language. Scala runs on the JVM. Java and Scala classes can be freely mixed, no matter whether they reside in different projects or in the same. Looking at the mechanics of how Scala code interacts with Openstack Swift object storage should help guide you to craft a Java equivalent.

From the above notebook, here are some steps illustrating how to configure and extract data from an Openstack Swift Object Storage instance using the Stocator library using the Scala language. The swift url decomposes into:

swift2d :// container . myacct / filename.extension
  ^            ^          ^            ^
stocator     name of   namespace    object storage
protocol     container               filename

Imports

import org.apache.spark.SparkContext
import scala.util.control.NonFatal
import play.api.libs.json.Json

val sqlctx = new SQLContext(sc)
val scplain = sqlctx.sparkContext

Sample Creds

// @hidden_cell
var credentials = scala.collection.mutable.HashMap[String, String](
  "auth_url"->"https://identity.open.softlayer.com",
  "project"->"object_storage_3xxxxxx3_xxxx_xxxx_xxxx_xxxxxxxxxxxx",
  "project_id"->"6xxxxxxxxxx04fxxxxxxxxxx6xxxxxx7",
  "region"->"dallas",
  "user_id"->"cxxxxxxxxxxaxxxxxxxxxx1xxxxxxxxx",
  "domain_id"->"cxxxxxxxxxxaxxyyyyyyxx1xxxxxxxxx",
  "domain_name"->"853255",
  "username"->"Admin_cxxxxxxxxxxaxxxxxxxxxx1xxxxxxxxx",
  "password"->"""&M7372!FAKE""",
  "container"->"notebooks",
  "tenantId"->"undefined",
  "filename"->"file.xml"
)

Helper Method

def setRemoteObjectStorageConfig(name:String, sc: SparkContext, dsConfiguration:String) : Boolean = {
    try {
        val result = scala.util.parsing.json.JSON.parseFull(dsConfiguration)
        result match {
            case Some(e:Map[String,String]) => {
                val prefix = "fs.swift2d.service." + name
                val hconf = sc.hadoopConfiguration
                hconf.set("fs.swift2d.impl","com.ibm.stocator.fs.ObjectStoreFileSystem")
                hconf.set(prefix + ".auth.url", e("auth_url") + "/v3/auth/tokens")
                hconf.set(prefix + ".tenant", e("project_id"))
                hconf.set(prefix + ".username", e("user_id"))
                hconf.set(prefix + ".password", e("password"))
                hconf.set(prefix + "auth.method", "keystoneV3")
                hconf.set(prefix + ".region", e("region"))
                hconf.setBoolean(prefix + ".public", true)
                println("Successfully modified sparkcontext object with remote Object Storage Credentials using datasource name " + name)
                println("")
                return true
            }
            case None => println("Failed.")
                return false
        }
    }
    catch {
       case NonFatal(exc) => println(exc)
           return false
    }
}

Load the Data

val setObjStor = setRemoteObjectStorageConfig("sparksql", scplain, Json.toJson(credentials.toMap).toString)
val data_rdd = scplain.textFile("swift2d://notebooks.sparksql/" + credentials("filename"))
data_rdd.take(5)