Quicksight is currently the best BI tool in my opinion. In the past I’ve chosen to use R/Python to make my apps, visuals, and dashboards. Due to the fact that I can customize them however I need and usually can figure a way to get better performance than any BI tool. Alongside being able to make custom templates and parametrized visuals / dashboards.
With Quicksight + SPICE that is no longer the case. It’s highly performant, offers a great API, with the agility to parameterize visuals / dashboards. I think to continue to make this the perfect product it needs to offer not only dashboards as an API but also SPICE datasets as an API too. Having the ability to create rest endpoints similar to as you would with a dashboard with parameters to query datasets would make this tool amazing. It’s a bit of a reach, but I think it is needed. Looking at other data science / analytics products like Posit Connect. Posit over charges the hell out of their connect product, but it allows you host shiny apps / dashboards. With Quicksight you can’t create as custom as an app / dashboard but 90% of what you would want to do is doable, much faster. You can schedule data updates, emails…etc with Posit, same you can do with Quicksight. From an auth standpoint both offer great security around datasets / apps with Quicksight being even better and easier in my opinion with row / column level security. The last thing something like Posit offers that I think is essential is datasets as endpoints via plumber. If Quicksight could offer some way to query spice datasets or host spice datasets as endpoints it would without a doubt be a huge feature. That being said SPICE is a special database which is focused on producing outputs for Quicksight and Quicksight only. I imagine there’s a way this can be adjusted though to just return raw queried data as an endpoint.
Hi @Sean_B - Welcome to AWS QuickSight Community and thanks for the feedback. Yes QuickSight is a great product and lots of good useful functionality are there, now coming to your queries whether SPICE data set can be present as a API end point, not sure but we can always highlight this to our QuickSight experts.
Tagging @Jesse@Max@n_vetri for their feedback on this and this something in feature list or not.
Hi @Kristin - Find a good feedback on QuickSight product
@Sean_B thanks for the feedback! If I understand your request correctly, you are hoping to be able to query a SPICE dataset programmatically. Is that correct? We do not have this on our current roadmap but we can pass your feedback to our product team. Can you help us understand what this would enable you to do? What types of use cases do you have for querying the data in SPICE directly using another tool (is it that you want to use open source visualizations on data in SPICE, or you want to build some sort of alerting workflow off the data in SPICE, something else)?
@Jesse So for me personally this would be a huge addition for my own business, however my business is more of using Quicksight as a public facing tool vs most others probably using it as an internal analytics tool. Specifically I pretty much embed all of Quicksight into a basic web app which users can search for visuals using AWS Cloudsearch. Due to Quicksight already having this great parameterized abilities as restful endpoints this allows me to create millions of great visuals with only a handful of dashboards and minimal coding. Users like most of these visuals and even have the option to create their own if they sign up to be an author. However for some cases users just want the data, which they can get by manually clicking on the download csv/excel file. Those users are usually using some other tool though like R/Python if they are doing that. It’d be so much easier if instead they could just hit an endpoint. A simple way I envision it is to allow programmatically downloading the csv/excel files using the current parameter ecosystem. So say I have a dashboard with this url https://us-east-2.quicksight.aws.amazon.com/sn/analyses/11111111111#p.id1=AAPL, if a user could just download the csv/excel file by adding an argument like download https://us-east-2.quicksight.aws.amazon.com/sn/analyses/11111111111#p.id1=AAPL&download=TRUE. Which then just initiates the file download right away, allowing me to ping that URL using R/Python and fetch the data. This also essentially creates a really easy and user friendly API builder for non technical users.
So for me personally it would eliminate the need for me creating a whole backend rest API for users to query data. Quicksight can do the vast majority of the visuals / analyses / adhoc requests. However there are some edge cases where users just want the data, which luckily Quicksight already offers csv/excel downloads of data, but also offering restful endpoints without the manual need to click download would be so useful.
Even with how most businesses probably already use Quicksight as an internal tool this would be a huge feature. If Quicksight is acting as maybe not the primary source but the curated data layer for all of a businesses reporting / analytical needs, then all data should ideally come from there. Some people might want to use R/Python or whatever tool to create their own custom app, report, ml model…etc. I cannot tell you how many times I would need to manually click download on various different BI tools in previous corporate roles (looker/datastudio/powerbi…etc).
@Sean_B I see, thanks for the detailed explanation! While an API/endpoint for SPICE datasets is not on the current roadmap, stay tuned for an API to retrieve CSV data behind table and pivot table visuals (will be part of the Paginated Reporting feature add-on, FYI). Should be able to build a call to that API into your workflows.
