Scaling up distributed serverless systems comes with its own set of engineering roadblocks and problems, and the fact that the concept of serverless systems is still in a very infantile stage, means that most of those problems are still unsolved. But, that shouldn't stop us from trying to solve and work around these roadblocks.
We will try and understand some of these roadblocks, and also learn how to solve or work around them, as discussed here:
- This is more of an architect's mistake rather than a roadblock. However, it is important to address this as one too many architects/software engineers fell and fall into the overestimation or the underestimation trap. The problem we will try to address is the exact number of instances you have to launch when scaling up. In most self-hosted MapReduce-style systems, it is taken...