Mews

Makes PCI-DSS level 1 compliance much simpler and minimizes the attack surface of our system. The sensitive credit card data don't even touch our infrastructure which eliminates lot of items from the long list of PCI-DSS requirements.

Power BI is really easy to start with. If you have just several Excel sheets or CSV files, or you build your first automated pipeline, it is actually quite intuitive to build your first reports.

And as we have kept growing, all the additional features and tools were just there within the Azure platform and/or Office 365.

Since we started building Mews, we have already passed several milestones in becoming start up, later also a scale up company and now getting ready to grow even further, and during all these phases Power BI was just the right tool for us.

Azure Data Factory was a great fit into our current stack where we already rely heavily on the Azure platform. We already had several Azure SQL Databases, so when we needed to optimize and start aggregating some of the historical data for our BI reporting in Power BI, we could use ADF to easily design and schedule the necessary pipelines.

Microsoft has massive expertise in writing development IDEs and it's really visible. Every since they released Visual Studio Code we benefit from its constant improvements. I can't imagine writing code in anything else anymore.

Writing code without static analysis tool is like driving blind. We choosed ESLint as it's clearly state of the art on frontend and supported by large ecosystem of plugins for various specific problems.

This was no real choice - we switched the moment Yarn was available, and never looked back. Yarn is the only reasonable frontend package manager that's actually being developed. They even aim to heal the node_modules madness with v2! Npm is just copying its ideas on top of introducing massive bugs with every change.

I'm not sure when it happened, but TypeScript is the new cool kid around the block. And we can see the massive potential it's unlocking when included in large scale codebase like ours. Types saves lot of mistakes, questions and searches for documentations, and we're happy to have them.

Have you ever tried to develop and debug large scale 2-way data binding? We did. And that's why we were happy to adopt single-way dataflow pattern that is central to the Redux.

We started with React when it was still in early stages (I think around version 0.13). Back then, it was major step up over plain HTML templates that just weren't debuggable at all. Together with emerging Redux (Flux in that time) pattern it took frontend by storm.

It provides "Error reporting" as a service. Allows aggreate errors nicely across multiple application instances. Our house-build solution simply was not enough at some point.