• Contents:

Amazon QuickSight Generic Overview

Ref: https://www.udemy.com/course/aws-certified-machine-learning-engineer-associate-mla-c01/learn/lecture/45287043

• 🔧 Serverless data analytics & visualization service
  • Quickly and easily create dashboards, graphs, charts, and reports from stored data
    • Fast and easy visualizations, ad-hoc analysis, business insights…
  • Can be used anytime, on any device (browsers/mobile) → no need to use AWS UI or CLI
  • 💡 More geared towards end users (business people) instead of tech/dev people
• Many data sources supported
  • DBs: Redshift, RDS, Aurora, EC2-hosted DBs (if they follow standards)…
  • Athena
  • Data stores: S3 data lakes, external DWHs…
    • Files: Excel, CSV, TSV, common/extended log format
  • AWS IoT Analytics
  • SaaS apps (e.g. Salesforce)
  • Any JDBC/ODBC data source (either cloud or on-premises!)
  • ❗ Can even perform simple/limited ETL on data to prepare it for visualization
• Use cases
  • Interactive & ad-hoc exploration/visualization of data
    • 💡 Slice and dice data on the fly to gain insights
  • Dashboards & KPIs
    • e.g. paginated reports
  • Machine Learning Insights
    • Anomaly detection (uses Random Cut Forest under the hood)
    • Forecasting (takes into account seasonalities)
    • Auto-narratives → Build rich dashboards with embedded narratives (tell the story of your data)
      • Screenshot
    • ❗ …and that, together with QuickSight Q, are all the ML capabilities of QuickSight so far! Nothing else yet!
    • 💡 These can all be done by non-tech users, very handy and easy to use!
    • Screenshot
• ❗ Anti-patterns
  • ETL → Use Glue instead for ETL-intensive workloads (QuickSight only does limited ETL)
• Security
  • Offers MFA on your account
  • VPC connectivity
    • Add QuickSight's IP range to SGs of your DB instances, can then access data
    • If private VPC access desired → ENI or DX
  • Granularity: Row-level security (RLS)
    • Column-level security (CLS) also supported in Enterprise Edition
• User Management → QuickSight is a web-app that non-tech users can access
  • ❗ Billed per user → you don't want anyone accessing the dashboards!
  • Users can be defined as IAM users, email signup also possible
  • SAML-based single sign-on (SSO) → uses IAM under the hood
  • MS Active Directory integration (Enterprise Edition)
  • Everything has MFA support

SPICE

• 🔧 QuickSight uses SPICE under the hood
  • SPICE = Super-fast, Parallel, In-memory, Calculation Engine
  • Datasets imported into SPICE, visualizations produced very fast
• Uses columnar storage & in-memory machine-code generation
  • Accelerates interactive queries on large datasets
  • In-memory calculation limited to 10GB of SPICE per user (limit can be increased at additional cost)
  • Highly available/durable, scalable to many users (100s of 1000s)

QuickSight Q

• 🔧 Answer BI questions with NLP
  • e.g. “What are the top-selling items in Florida?”
    • 👍 Advantage: No need to figure out SQL or drag & drop into chart
  • Powered by Amazon Q (ML-powered)
• Caveats
  • Offered as add-on in certain regions
  • Costs extra
  • Requires personal training on how to use
  • ❗ Doesn't work out-of-the-box → must set up topics associated with datasets
    • Datasets & fields must be NLP-friendly
    • Date-handling must be defined

Paginated Reports

• 🔧 Highly formatted, multi-page reports designed to be printed for management
• May span many pages, can be built on top of existing QuickSight dashboards