Real Time Sales - In 100ms from sales to dashboard
What the episode covers
A tradition that underlines our focus on data. In this sales season, everybody in bol.com wants to know how we are doing, preferably Real-Time. With the latest version of the Real-Time Sales dashboard, our guests managed to get the information within 100ms from sales to the dashboard.
If you are one of the people that spent maybe a little too much time watching the results of the election in the US of A unfold on CNN or another network, you could catch the bug of this tradition too. We have a magic board, but we don’t have the talking heads that came with it....
In this episode, we talk about that one screen all bol.com colleagues have open all day in this season period. The initial version which was born back in 2004 and got the name Ancorama. Referring to the colleague who built it. It looked like an excel sheet being updated by querying the database. A couple of years ago a new initiative was born, Real-Time Sales. It brought great new graphs. But in fact, it still wasn't Real-Time.Now we are dealing with a brand new version and it is also known as Real-Time Sales, the next generation. Our guests had a huge stake in building this version and we want to know everything about it.
- Why and when did you start the project?
- How did you achieve 100ms processing time?
- What building blocks used and why?
- RSocket, R2DBC, Reactive programming, Vue3, Kotlin Coroutines
- Test approach
- Future developments
- Goos van den Bekerom; Software Engineer in the buying domain
- Marvin Zwolsman; Software Engineer in the buying domain and friend of the show! (Hey Google)
- Eugene Figueiras; Software Engineer in the Selling, Assortment and Fintech domain.
Blogpost on peak season frontend performance
Podcast on season preparations