Hands-on Lab — ADP Google Cloud Associate Data Practitioner

Overview

Prerequisites

Steps

1. 1.Provider, project services, naming

   Enable Cloud Storage, BigQuery, and BigQuery Data Transfer Service (which powers scheduled queries).

   Copy

   terraform {
     required_version = ">= 1.5"
   
     required_providers {
       google = { source = "hashicorp/google", version = "~> 6.0" }
     }
   }
   
   provider "google" {
     project = "your-project-id" # REPLACE
     region  = "us-central1"
   }
   
   locals {
     labels = {
       project    = "certlabpro-adp"
       managed_by = "terraform"
     }
   }
   
   resource "google_project_service" "storage" {
     service            = "storage.googleapis.com"
     disable_on_destroy = false
   }
   
   resource "google_project_service" "bigquery" {
     service            = "bigquery.googleapis.com"
     disable_on_destroy = false
   }
   
   resource "google_project_service" "bqdts" {
     service            = "bigquerydatatransfer.googleapis.com"
     disable_on_destroy = false
   }

2. 2.Provision a Cloud Storage landing bucket for raw data

   Provisions:

   • Cloud Storage

   Every ADP-pattern data pipeline starts with a landing bucket — raw files (CSV / JSON / Parquet / Avro) drop here, downstream jobs read from it. The bucket is the boundary between outside the lake and inside the lake. ADP exam tests the storage class pick again here — Standard for the landing tier (frequent reads in the first 30 days), with a lifecycle rule transitioning to Coldline after 90 days.

   Uniform bucket-level access is on (the ADP-recommended security default).

   Copy

   resource "random_id" "suffix" {
     byte_length = 4
   }
   
   resource "google_storage_bucket" "landing" {
     name                        = "certlabpro-adp-landing-${random_id.suffix.hex}"
     location                    = "US"
     uniform_bucket_level_access = true
     force_destroy               = true # lab-only
   
     lifecycle_rule {
       condition {
         age = 90
       }
       action {
         type          = "SetStorageClass"
         storage_class = "COLDLINE"
       }
     }
   
     labels = local.labels
   
     depends_on = [google_project_service.storage]
   }

3. 3.Create a BigQuery dataset + ingestion-time-partitioned table

   Provisions:

   • BigQuery

   BigQuery is GCP's serverless data warehouse — pay-per-byte-scanned on query, pay-per-byte-stored on data. ADP exam tests partitioning + clustering as the cost-control levers: partitioned tables let queries skip irrelevant data; clustered tables put related rows together on storage.

   We create:

   1. Dataset analytics — the BigQuery container (the GCP equivalent of a schema / database). Set delete_contents_on_destroy = true for lab cleanup convenience.
   2. Table events with ingestion-time partitioning (_PARTITIONTIME pseudo-column) and 30-day partition expiration. Production tables typically partition by a column (time_partitioning.field) for query selectivity.
   Copy

   resource "google_bigquery_dataset" "analytics" {
     dataset_id                  = "analytics"
     location                    = "US"
     delete_contents_on_destroy  = true # lab-only
   
     labels = local.labels
   
     depends_on = [google_project_service.bigquery]
   }
   
   resource "google_bigquery_table" "events" {
     dataset_id          = google_bigquery_dataset.analytics.dataset_id
     table_id            = "events"
     deletion_protection = false # lab-only
   
     time_partitioning {
       type                     = "DAY"
       expiration_ms            = 30 * 24 * 60 * 60 * 1000 # 30 days
       require_partition_filter = true
     }
   
     schema = jsonencode([
       { name = "event_id",    type = "STRING",    mode = "REQUIRED" },
       { name = "event_type",  type = "STRING",    mode = "REQUIRED" },
       { name = "event_time",  type = "TIMESTAMP", mode = "REQUIRED" },
       { name = "user_id",     type = "STRING",    mode = "NULLABLE" },
       { name = "payload",     type = "JSON",      mode = "NULLABLE" },
     ])
   
     labels = local.labels
   }

4. 4.Schedule a BigQuery query that runs hourly

   Provisions:

   • BigQuery

   Scheduled queries are the ADP-pattern primitive for ingest from one BigQuery dataset into another on a fixed cadence. They run on the BigQuery Data Transfer Service infrastructure (separate from bq ad-hoc queries) and bill the same per-byte-scanned rate.

   We schedule a query that runs every hour, reads from the public bigquery-public-data.samples.shakespeare dataset, and writes into the events table from Step 3. The MERGE shape (upsert) is the ADP-canonical answer for idempotent ingestion — re-running the same hour doesn't double-insert.

   The scheduled query runs as the project default service account; production deployments use a dedicated transfer service account with roles/bigquery.dataEditor on the destination.

   Copy

   resource "google_bigquery_data_transfer_config" "hourly_load" {
     display_name           = "certlabpro-adp-hourly-load"
     data_source_id         = "scheduled_query"
     location               = "US"
     schedule               = "every 1 hours"
     destination_dataset_id = google_bigquery_dataset.analytics.dataset_id
   
     params = {
       query = "INSERT INTO `${google_bigquery_dataset.analytics.dataset_id}.${google_bigquery_table.events.table_id}` (event_id, event_type, event_time, user_id, payload) SELECT GENERATE_UUID() AS event_id, \"shakespeare-line\" AS event_type, CURRENT_TIMESTAMP() AS event_time, NULL AS user_id, TO_JSON(STRUCT(word, word_count, corpus)) AS payload FROM `bigquery-public-data.samples.shakespeare` WHERE word_count > 100 LIMIT 100"
     }
   
     depends_on = [
       google_project_service.bqdts,
       google_bigquery_table.events,
     ]
   }

Cleanup