AWS QuickSight

Ben Lee
December 8, 2016

Last month Amazon Web Services (AWS) have made their Business Intelligence (BI) solution QuickSight generally available for all customers. They describe it like this:

“QuickSight is a fast, cloud-powered business analytics service that makes it easy for you to build visualizations, perform ad-hoc analysis, and quickly get business insights from your data.”

I had an opportunity this week to try it out for a proof of concept. Like all things AWS, you can get to it from the AWS Console.

At the time of writing this product is only available in N Virginia, Oregon and Ireland. I’m sure like other AWS products it will eventually be released everywhere.
Like all good users, I jumped in without reading the documentation properly, below are my initial observations:

  • There is reasonable support for data sources at the moment. Biggest omissions are; no Oracle, JSON and XML. The supported data sources are below:
  • Amazon Redshift
  • Amazon Aurora (Amazon RDS only)
  • MariaDB 10.0 or later
  • Microsoft SQL Server 2012 or later
  • MySQL 5.1 or later
  • PostgreSQL 9.3.1 or later
  • Microsoft Excel files (.xlsx files) on your local network
  • Delimited text files (for example, .csv and .tsv files) on your local network or in Amazon S3
  • Common log format (.clf) and extended log format (.elf) files on your local network or in Amazon S3
  • Salesforce
  • The data limits currently imposed on QuickSight should not cause issues in most situations, they are:
  • File imports limited to 1GB
  • Files can have up to 200 columns, with up to 25400 characters per row
  • Field length limitation of 511 characters
  • Max DB table import limit 10GB
  • Supported data types are below, one interesting point, however, is that if the data for a field is invalid then it is omitted from the data loaded into memory, for example:
  • Date (In a supported format)
  • Decimal (Only 4 decimal places is supported the rest truncated)
  • Integer
  • String
  • User & Access Management is very basic. Everything is done on a per-user basis using IAM users.
  • The UI needs a lot more polish, for example:
  • It lacks the intuitive feel a lot of other tools seem to have.
  • It feels like the drive to keep the UI simple has actually complicated things.
  • Tooltips would be handy to improve the experience.
  • There were general bugs in the UI. A couple examples below
  • I could not see values that I had converted to a measure as the allocated space to display them was too small
  • When previewing data if you have more columns than could fit on a page there is no obvious way to view what fit on the screen
  • Graph Customisation options are limited
  • Data Filter abilities are limited
  • You can not join across different data sources
  • You can connect to remote databases not hosted by AWS which is a very rare feature for a cloud offering
  • There is capability to create dashboards

Based on the observations above, AWS still have a lot of work before I would consider it for a real project.
All the code, all the fun – Ben

Ben writes blogs on the technical side of BI, the code, all the code and not much other than the code.

Connect with Ben on LinkedIn or read some of his other blogs here.

Copyright © 2019 OptimalBI LTD.