instructions-azure-to-s3-export.md

Loading SQL Server tables to Spark on EC2 and output to S3

0. Setup sqlserver on Azure

Follow the instructions here https://azure.microsoft.com/en-us/documentation/articles/sql-database-get-started/ to setup a SQL Server instance on Azure

1. Launch spark cluster using spark-ec2 script

Export AWS credentials

export AWS_ACCESS_KEY_ID=your_aws_key
export AWS_SECRET_ACCESS_KEY=your_aws_secret
export AWS_PRIVATE_KEY=your_aws_private_key

Clone the spark repo and cd to the ec2 sub-directory

git clone https://github.com/apache/spark.git

cd spark

./spark-ec2 --key-pair=$AWS_PRIVATE_KEY --identity-file=$AWS_PRIVATE_KEY.pem --region=us-east-1 --zone=us-east-1a  --spark-version=1.5.1 launch my-spark-cluster

ssh to the cluster

./spark-ec2 -k $AWS_PRIVATE_KEY -i $AWS_PRIVATE_KEY.pem login my-spark-cluster

2. Set up the master instance for development

Update aws-cli, set up Java 8 and maven

yum install -y aws-cli java-1.8.0-openjdk java-1.8.0-openjdk-devel
export JAVA_HOME=/usr/lib/jvm/java-1.8.0
export PATH=$JAVA_HOME/bin:$PATH

wget http://apache.mirrors.hoobly.com//maven/maven-3/3.3.3/binaries/apache-maven-3.3.3-bin.tar.gz
tar xvfz apache-maven-3.3.3-bin.tar.gz
sudo mv apache-maven-3.3.3 /opt/maven
sudo ln -s /opt/maven/bin/mvn /usr/bin/mvn

Export AWS and Azure credentials

export SQLSERVER_USERNAME=<your SQL Sever user name>
export SQLSERVER_PWD=<your SQL Server password>
export SQLSERVER_HOST_PORT=<your SQL Server hostname and port, e.g (mine is) ngc1379w4v.database.windows.net:1433>
export SQLSERVER_DB=<your SQL Server database>
export SQLSERVER_TABLE=<your SQL Server table>
export AWS_ACCESS_KEY_ID=<your AWS access key>
export AWS_SECRET_ACCESS_KEY=<your AWS secret>
export AWS_S3_BUCKET=<s3 bucket to store output, e.g tomzeng/rdd-output>

Clone the sample app repo

git clone https://github.com/tomz/SparkApps.git

Change into the sample app directory

cd SparkApps

copy the SQL Server jdbc driver

aws s3 cp s3://tomzeng/jdbc-drivers/sqljdbc4.jar .

3. Build and run the app

Build the app

mvn install

Run the app

~/spark/bin/spark-submit --master local[*] --class com.sparkexpert.MainMS --jars sqljdbc4.jar --packages com.databricks:spark-csv_2.11:1.2.0 spark-load-from-db/target/spark-load-from-db-1.0-SNAPSHOT.jar

4. Check your s3 bucket for the output

aws s3 ls s3://$AWS_S3_BUCKET/