How we organized a highly effective and cost-efficient DataLake and why we did it this way
We live in an amazing time where you can quickly and easily connect several ready-to-use open tools, configure them with 'disconnected awareness' following advice from Stack Overflow, without delving into the 'many letters', and launch them into commercial operation. But when it's time to update/expand or someone accidentally reboots a couple of machines, you realize that you've entered some persistent, annoying nightmare awake, and everything has suddenly become much more complicated.
