Use Presto with Cloud Dataproc

Presto is a distributed SQL query engine designed to query large data sets distributed over one or more heterogeneous data sources. Presto can query Hive, MySQL, Kafka and other data sources through connectors. This tutorial shows you how to:

Install the Presto service on a Cloud Dataproc cluster
Query public data from a Presto client installed on your local machine that communicates with a Presto service on your cluster
Run queries from a Java application that communicates with the Presto service on your cluster through the Presto Java JDBC driver.

Objectives

Create a Cloud Dataproc cluster with Presto installed
Prepare data. This tutorial uses the Chicago Taxi Trips public dataset, available in BigQuery.
1. Extract the data from BigQuery
2. Load the data into Cloud Storage as CSV files
3. Transform data:
  1. Expose the data as a Hive external table to make the data queryable by Presto
  2. Convert the data from CSV format into Parquet format to make querying faster
Send Presto CLI or application code queries using an SSH tunnel or Presto JDBC driver, respectively, to the Presto coordinator running on the cluster
Check logs and monitor the Presto service through the Presto Web UI

Costs

This tutorial uses billable components of Google Cloud Platform, including:

Use the Pricing Calculator to generate a cost estimate based on your projected usage. New Google Cloud Platform users may be eligible for a free trial.

Before you begin

If you haven't already done so, create a Google Cloud Platform project and a Cloud Storage bucket to hold the data used in this tutorial.

Setting up your project
1. Google アカウントにログインします。
  Google アカウントをまだお持ちでない場合は、新しいアカウントを登録します。
2. GCP プロジェクトを選択または作成します。
  
  注: この手順で作成するリソースをそのまま保持する予定でない場合、既存のプロジェクトを選択するのではなく、プロジェクトを作成してください。チュートリアルの完了後にそのプロジェクトを削除すれば、プロジェクトに関連するすべてのリソースを削除できます。
  
  プロジェクトセレクタのページに移動
3. Google Cloud Platform プロジェクトに対して課金が有効になっていることを確認します。詳しくは、課金を有効にする方法をご覧ください。
4. Cloud Dataproc, Compute Engine, Cloud Storage, and BigQuery API を有効にします。
  APIを有効にする
5. Cloud SDK をインストールして初期化します。
Creating a Cloud Storage bucket in your project to hold the data used in this tutorial.
1. GCP Console で、Cloud Storage ブラウザページに移動します。
  Cloud Storage ブラウザページに移動
2. [バケットを作成] をクリックします。
3. [バケットを作成] ダイアログ内で、以下の属性を指定します。
  - バケット名の要件を満たす、一意のバケット名。
  - ストレージクラス.
  - バケットデータが保存されるロケーション。
4. [作成] をクリックします。

Install the Presto client

Install the Presto client software on your local machine.

Download the presto-cli-nnn-executable.jar to your local machine, rename it to presto-cli, and make it executable.
```
chmod a+x presto-cli
```

Create a Cloud Dataproc cluster

Optional Components^Beta Update: You can now ask Cloud Dataproc to install Presto on your cluster without using an initialization action (see Optional Components→Presto). To do this, change the gcloud dataproc clusters create command, below, to gcloud beta dataproc clusters create, and replace the --initialization-actions flag with --optional-components=PRESTO and --image-version=1.3-deb9 flags. If using the optional component, the Presto server will run on port 8060.

Create a Cloud Dataproc cluster using an initialization action to install the Presto service software on the cluster.

Set environment variables:
- PROJECT: your project ID
- BUCKET_NAME: the name of the Cloud Storage bucket you created in Before you begin
- REGION: region where the cluster used in this tutorial will be created, for example, "us-west1"
- ZONE: zone within the cluster's region where the cluster used in this tutorial will be located, for example, "us-west1-a"
- WORKERS: 3 - 5 workers are recommended for this tutorial
```
export PROJECT=project-id
export WORKERS=number
export REGION=region
export ZONE=zone
export BUCKET_NAME=bucket-name
```
Create a cluster using the default or a custom initialization script.
Default
1. Set a local INIT_ACTION variable to point to the location of the initialization script in Cloud Storage that will install the Presto service on a new Cloud Dataproc cluster.
```
export INIT_ACTION=gs://dataproc-initialization-actions/presto/presto.sh
```
2. Run the gcloud command-line tool on your local machine to create the cluster.
```
gcloud dataproc clusters create presto-cluster \
    --project=${PROJECT} \
    --zone=${ZONE} \
    --num-workers=${WORKERS} \
    --scopes=cloud-platform \
    --initialization-actions=${INIT_ACTION}
```
  The Presto initialization action, which is co-located in the GitHub dataproc-initialization-actions repository, installs Presto at on your cluster's master node at /presto-server-version.
  
  Creating a High-Availability cluster – a cluster with multiple master nodes – is not recommended since the coordinator is started on master node 0, and any other other master nodes will remain idle.
Custom
1. View the GitHub Presto initialization script to determine the initialization settings to modify. The initialization script settings include the following:
  - Presto version (see the GitHub Presto initialization script) for the currently set default version)
  - Presto HTTP port (default: 8080)
  - Connector (default: Apache Hive connector)
2. Copy the Presto initialization action from Cloud Storage to your local machine.
```
gsutil cp gs://dataproc-initialization-actions/presto/presto.sh local-path
```
3. Modify the initialization script to customize the Presto configuration.
4. Upload the custom Presto initialization action from your local machine to the Cloud Storage bucket you created in Before you begin.
```
gsutil cp presto.sh gs://${BUCKET_NAME}/
```
5. Set a local INIT_ACTION variable to point to the location of your custom script in Cloud Storage
```
export INIT_ACTION=gs://${BUCKET_NAME}/presto.sh
```
6. Run the gcloud command-line tool on your local machine to create the cluster.
```
gcloud dataproc clusters create presto-cluster \
    --project=${PROJECT} \
    --zone=${ZONE} \
    --num-workers=${WORKERS} \
    --scopes=cloud-platform \
    --initialization-actions=${INIT_ACTION}
```
  Creating a High-Availability cluster – a cluster with multiple master nodes – is not recommended since the coordinator is started on master node 0, and other master nodes will remain idle.
To monitor or troubleshoot the Presto installation on your cluster, SSH into cluster VMs, then view the initialization actions logs.
If an error is reported that indicates the initialization action failed, for example due to a timeout error, delete the cluster and re-run the gcloud dataproc clusters create command to re-run the initialization script on a new cluster.

Prepare data

Export the bigquery-public-data chicago_taxi_trips dataset to Cloud Storage as CSV files, then create a Hive external table to reference the data.

On your local machine, run the following command to import the taxi data from BigQuery as CSV files without headers into the Cloud Storage bucket you created in Before you begin.

bq --location=us extract --destination_format=CSV \
     --field_delimiter=',' --print_header=false \
       "bigquery-public-data:chicago_taxi_trips.taxi_trips" \
       gs://${BUCKET_NAME}/chicago_taxi_trips/csv/shard-*.csv

Create Hive external tables that are backed by the CSV and Parquet files in your Cloud Storage bucket.

Create the Hive external table chicago_taxi_trips_csv.

gcloud dataproc jobs submit hive \
    --cluster presto-cluster \
    --execute "
        CREATE EXTERNAL TABLE chicago_taxi_trips_csv(
          unique_key   STRING,
          taxi_id  STRING,
          trip_start_timestamp  TIMESTAMP,
          trip_end_timestamp  TIMESTAMP,
          trip_seconds  INT,
          trip_miles   FLOAT,
          pickup_census_tract  INT,
          dropoff_census_tract  INT,
          pickup_community_area  INT,
          dropoff_community_area  INT,
          fare  FLOAT,
          tips  FLOAT,
          tolls  FLOAT,
          extras  FLOAT,
          trip_total  FLOAT,
          payment_type  STRING,
          company  STRING,
          pickup_latitude  FLOAT,
          pickup_longitude  FLOAT,
          pickup_location  STRING,
          dropoff_latitude  FLOAT,
          dropoff_longitude  FLOAT,
          dropoff_location  STRING)
        ROW FORMAT DELIMITED
        FIELDS TERMINATED BY ','
        STORED AS TEXTFILE
        location 'gs://${BUCKET_NAME}/chicago_taxi_trips/csv/';"

Verify the creation of the Hive external table.
Skip this step to save time and resource usage. If your cluster has 3 or fewer nodes, this query can take several minutes. You can skip this query step and only verify the creation of the final Hive Parquet, below, since that query step is significantly faster.
```
gcloud dataproc jobs submit hive \
    --cluster presto-cluster \
    --execute "SELECT COUNT(*) FROM chicago_taxi_trips_csv;"
```

Create another Hive external table chicago_taxi_trips_parquet with the same columns, but with data stored in Parquetformat for better query performance.

gcloud dataproc jobs submit hive \
    --cluster presto-cluster \
    --execute "
        CREATE EXTERNAL TABLE chicago_taxi_trips_parquet(
          unique_key   STRING,
          taxi_id  STRING,
          trip_start_timestamp  TIMESTAMP,
          trip_end_timestamp  TIMESTAMP,
          trip_seconds  INT,
          trip_miles   FLOAT,
          pickup_census_tract  INT,
          dropoff_census_tract  INT,
          pickup_community_area  INT,
          dropoff_community_area  INT,
          fare  FLOAT,
          tips  FLOAT,
          tolls  FLOAT,
          extras  FLOAT,
          trip_total  FLOAT,
          payment_type  STRING,
          company  STRING,
          pickup_latitude  FLOAT,
          pickup_longitude  FLOAT,
          pickup_location  STRING,
          dropoff_latitude  FLOAT,
          dropoff_longitude  FLOAT,
          dropoff_location  STRING)
        STORED AS PARQUET
        location 'gs://${BUCKET_NAME}/chicago_taxi_trips/parquet/';"

Load the data from the Hive CSV table into the Hive Parquet table.

gcloud dataproc jobs submit hive \
    --cluster presto-cluster \
    --execute "
        INSERT OVERWRITE TABLE chicago_taxi_trips_parquet
        SELECT * FROM chicago_taxi_trips_csv;"

Verify that the data loaded correctly.

gcloud dataproc jobs submit hive \
    --cluster presto-cluster \
    --execute "SELECT COUNT(*) FROM chicago_taxi_trips_parquet;"

Run queries

You can run queries locally from the Presto CLI or from an application.

Presto CLI queries

This section demonstrates how to query the Hive Parquet taxi dataset using the Presto CLI.

1. Create an SSH tunnel. Run the following command to use dynamic port forwarding to create a localhost:1080 SOCKS proxy on your local machine that connects through an SSH tunnel to the master node of the Cloud Dataproc cluster. The terminal will hang during the execution of the command.

gcloud compute ssh presto-cluster-m \
    --project=${PROJECT} \
    --zone=${ZONE} \
    -- -D 1080 -N

In a new terminal window on your local machine, run the Presto CLI and connect to the Presto server running on port 8080 of the master node of the cluster.
```
./presto-cli \
    --server presto-cluster-m:8080 \
    --socks-proxy localhost:1080 \
    --catalog hive \
    --schema default
```

At the presto:default> prompt, verify that Presto can find the Hive tables.

show tables;

       Table
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
chicago_taxi_trips_csv
chicago_taxi_trips_parquet
(2 rows)

Run queries from the presto:default> prompt, and compare the performance of querying Parquet versus CSV data.

Parquet data query

select count(*) from chicago_taxi_trips_parquet where trip_miles > 50;

_col0
‐‐‐‐‐‐‐‐
117957
(1 row)

Query 20180928_171735_00006_2sz8c, FINISHED, 3 nodes
Splits: 308 total, 308 done (100.00%)
0:16 [113M rows, 297MB] [6.91M rows/s, 18.2MB/s]

CSV data query

select count(*) from chicago_taxi_trips_csv where trip_miles > 50;

_col0
‐‐‐‐‐‐‐‐
117957
(1 row)

Query 20180928_171936_00009_2sz8c, FINISHED, 3 nodes
Splits: 881 total, 881 done (100.00%)
0:47 [113M rows, 41.5GB] [2.42M rows/s, 911MB/s]

Java application queries

To run queries from a Java application through the Presto Java JDBC driver:

Download the Presto Java JDBC driver.

Add a presto-jdbc dependency in Maven pom.xml.

<dependency>
<groupId>com.facebook.presto</groupId>
<artifactId>presto-jdbc</artifactId>
<version>0.206</version>
</dependency>

Sample Java code

package dataproc.codelab.presto;

import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.ResultSet;
import java.sql.SQLException;
import java.sql.Statement;
import java.util.Properties;

public class PrestoQuery {
  private static final String URL = "jdbc:presto://presto-cluster-m:8080/hive/default";
  private static final String SOCKS_PROXY = "localhost:1080";
  private static final String USER = "user";
  private static final String QUERY =
      "select count(*) as count from chicago_taxi_trips_parquet where trip_miles > 50";

  public static void main(String[] args) {
    try {
      Properties properties = new Properties();
      properties.setProperty("user", USER);
      properties.setProperty("socksProxy", SOCKS_PROXY);
      Connection connection = DriverManager.getConnection(URL, properties);
      try (Statement stmt = connection.createStatement()) {
        ResultSet rs = stmt.executeQuery(QUERY);
        while (rs.next()) {
          int count = rs.getInt("count");
          System.out.println("The number of long trips: " + count);
        }
      }
    } catch (SQLException e) {
      e.printStackTrace();
    }
  }
}

Logging and monitoring

Logging

By default, the Presto initialization action configures /var/presto/data as the Presto data directory on the cluster's master and worker nodes. The Presto logs are located at /var/presto/data/var/log/.

Web UI

By default, the Presto initialization action configures the Presto service to run on port 8080 of the cluster's master node. To connect to the Presto web interface on the master node:

Create an SSH tunnel with SOCKS proxy from you local machine to master-node port 1080.

Launch a Chrome browser with the socks5://localhost:1080 proxy.

Linux

/usr/bin/google-chrome \
    --proxy-server="socks5://localhost:1080" \
    --host-resolver-rules="MAP * 0.0.0.0 , EXCLUDE localhost" \
    --user-data-dir=/tmp/presto-cluster-m

macOS

"/Applications/Google Chrome.app/Contents/MacOS/Google Chrome" \
    --proxy-server="socks5://localhost:1080" \
    --host-resolver-rules="MAP * 0.0.0.0 , EXCLUDE localhost" \
    --user-data-dir=/tmp/presto-cluster-m

In the browser, go to:
```
http://presto-cluster-m:8080
```

Monitoring

Presto exposes cluster runtime information through runtime tables. In a Presto session (from the presto:default>) prompt, run the following query to view runtime table data:

select * FROM system.runtime.nodes;

Cleaning up

After you've finished the Use Presto with Cloud Dataproc tutorial, you can clean up the resources that you created on GCP so they won't take up quota and you won't be billed for them in the future. The following sections describe how to delete or turn off these resources.

Deleting the project

The easiest way to eliminate billing is to delete the project that you created for the tutorial.

To delete the project:

注意: プロジェクトを削除すると、次の影響があります。

プロジェクト内のすべてのものが削除されます。 このチュートリアルで既存のプロジェクトを使用した場合、それを削除すると、そのプロジェクトで行った他の作業もすべて削除されます。
カスタムプロジェクト ID は失われます。このプロジェクトを作成したときに、将来使用するカスタムプロジェクト ID を作成した可能性があります。appspot.com URL など、プロジェクト ID を使用する URL を保持するには、プロジェクト全体を削除するのではなく、プロジェクト内の選択したリソースを削除します。

複数のチュートリアルとクイックスタートを検討する予定がある場合は、プロジェクトを再利用すると、プロジェクトの割り当て制限を超えないようにすることができます。

GCP Console で [プロジェクト] ページに移動します。
プロジェクトページに移動
プロジェクトリストで、削除するプロジェクトを選択し、[削除] をクリックします。
ダイアログでプロジェクト ID を入力し、[シャットダウン] をクリックしてプロジェクトを削除します。

Deleting the cluster

To delete your cluster:

gcloud dataproc clusters delete --project=${PROJECT} presto-cluster

Deleting the bucket

To delete the Cloud Storage bucket you created in Before you begin, including the data files stored in the bucket:
```
gsutil -m rm -r gs://${BUCKET_NAME}
```