Context
Evolution of dataviz architecture
Challenge
Switch from a single-node tableau server architecture to a distributed architecture
Technologies used
Cloud AWS • Tableau Server 2020 • DBMS: SAP HANA 1.0 SPS12 & HANA 2.0 Full Use • ETL: SAP BW & HANA stored procedures • OS for Tableau Server: Windows 2019
Methodology and implementation
Migration from a single-node Tableau Server architecture to a distributed cluster on AWS:
**Target architecture** - 3 Tableau Server nodes on AWS EC2 (Windows Server 2019) - Load balancer for high availability - Direct connection to SAP HANA (ODBC) - External repository on PostgreSQL
**Sizing** - Current load analysis (users, views, extracts) - CPU/RAM requirements calculation per Tableau process - Choice of appropriate EC2 instances - Storage configuration (EBS)
**Migration** - Complete backup of single-node instance - Installation of first cluster node - Migration of workbooks and data sources - Addition of secondary nodes - Failover testing
**Optimizations** - Distribution of Tableau processes across nodes - Cache configuration - Extract optimization - CloudWatch monitoring and alerts
Results
Tableau Server Cluster 2020 architecture implemented and documented. Performance gain was evaluated at +59% based on a ratio sum(loads)/sum(elapsed time)
Visualizations
Tableau Server cluster architecture on AWS with HANA connection