Cloudera Data Platform Migration

On-prem big data infrastructure moved to Azure PaaS

Client: Oil & Gas Industry Leader, USA

16

months to full migration, against 18 planned

4

months to MVP

8+ TB

of data, 55+ ingestion components, 25+ analytics products

TypeScriptNode.jsPythonPostgreSQLRabbitMQAWSAmazon EC2AWS compute clusterAWS container servicesAmazon SQSAWS registryData centerAzure Kubernetes ServiceAzure Key VaultAzure DevOps servicesAzure Data Lake Storage Gen2Microsoft platformPower BICEPH S3-compatible storageTerraformPlaywrightSeleniumJasmineKarma
TypeScriptNode.jsPythonPostgreSQLRabbitMQAWSAmazon EC2AWS compute clusterAWS container servicesAmazon SQSAWS registryData centerAzure Kubernetes ServiceAzure Key VaultAzure DevOps servicesAzure Data Lake Storage Gen2Microsoft platformPower BICEPH S3-compatible storageTerraformPlaywrightSeleniumJasmineKarma

Challenge

A complex on-prem Cloudera platform had to reach Azure within 18 months, with minimal refactoring and no disruption to the business. It carried 55+ ingestion components, 25+ analytics products and 8+ TB of varied data, tied to legacy technologies like Oozie and HDFS.

Solution

A phased Azure PaaS migration run MVP-first, with reusable accelerators doing the repetitive work of moving ingestion components and analytics products. Hub-and-spoke architecture, event-based pipeline execution, and custom tooling for automation, validation and monitoring.

Implementation

  • 01Discovery and complexity assessment of the existing systems
  • 02Migration PoCs and MVP on Azure HDInsight, ADLS Gen2 and Data Factory
  • 03Accelerators for ingestion generation, pipeline migration, CI/CD and data validation
  • 04Event-driven architecture, with ExpressRoute for on-prem connectivity
  • 05Hub-and-spoke resource separation for cost and security isolation

Business impact

MVP delivered in 4 months, full migration in 16

The on-prem platform decommissioned

New capabilities along the way: granular billing, smart monitoring, on-demand compute, data lineage

Automated ingestion and pipeline migration, faster and more consistent than by hand

Technology stack

Languages & Tools
  • Python
  • PowerShell
Data & Processing
  • Azure HDInsight (Spark/HIQ)
  • Databricks
  • ADLS Gen2
  • SQL Azure
  • SQL DWH
Cloud Services
  • Azure Data Factory
  • Event Grid
  • Azure Functions
  • Key Vault
  • Azure Monitor
  • ExpressRoute
BI & DevOps
  • Power BI
  • Azure DevOps

Have a complex system to build or modernize?

Tell us what must change. We will bring the relevant domain and engineering leads into the conversation.

Explore all case studies