There’s no question that Andy Jassy’s presentation at the AWS re:Invent 2015’s first keynote address covered a lot of ground across the spectrum of cloud technology. But his introduction of Amazon’s new cloud-based business intelligence solution, Amazon QuickSight, captured a lot of attention.
QuickSight, by leveraging the integration of AWS data-oriented services, is built to reduce both the cost and complexity long associated with converting streams of data into useful business intelligence. In fact, Amazon claims that QuickSight can cut 90% off the cost of traditional solutions.
Part of the trick is capitalizing on the sheer volume of raw data your AWS-based operation might already have available. There’s a good chance that you’re already at least reasonably invested in storage (S3), data warehousing (DynamoDB, Redshift), streaming data (EMR, Kinesis), and big data analytics (Amazon Machine Learning): why not bring them all together to see what extra value you can access?
What might really help QuickSight take off, though, is the fact that Amazon has placed business intelligence squarely in the hands of the people who need the answers the most. You’ll get full functionality right from the browser interface without the need for data professionals (in fact, at this point there isn’t even a native API). Or, in Amazon’s own words:
Just log in, point to a data source, and create your first visualization in minutes
Slide stacks of visualizations – or “stories” as AWS calls them – can be saved and shared among colleagues, adding to the value of the insights.
QuickSight’s in-memory calculation engine (called SPICE, for “Super-fast Parallel In-memory Computation Engine“) can easily scale to smoothly deliver visualizations to hundreds of thousands of users in a particular organization.
Here are some notes based on the Q&A session:
- Using SPICE as the main storage repository for your computational data is NOT recommended (even though SPICE data are archived and never automatically deleted).
- You can mask data while generating reports, which will be a great feature for organizations with multiple security levels. Intelligent filters also allow you to open up access only to those parts of the dataset needed by a particular dashboard user.
- You can incrementally import new data.
- You can export XLS reports.
- For best performance, feel free to import denormalized data from your DBMS. SPICE will generate all the needed views/tables and compress the dataset (up to 10x compression).
- Use either Lambda functions or the dashboard to perform ETL! (Think: column or row level transformations using formulas and so on).
Still to come: support for streaming datasets for real-time and for GEO data, and ODBC/JDBC and Salesforce external data connections.
Explore the Analytics tools provided by AWS, including QuickSight, Elastic Map Reduce (EMR), Data Pipeline, Elasticsearch, Kinesis, Amazon Machine Learning in the Cloud Academy’s Analytics Fundamentals for AWS course.
