Traditional databases are unable to scale to meet the demands of global cloud applications because they were designed to scale vertically. Modern databases have been designed to leverage sharding, which allows them to scale horizontally by partitioning the data across multiple nodes. This improves responsiveness because it reduces contention for disk access; it provides resilience because the data is also stored redundantly across several machines; and allows the database to be elastic, as the demand grows, by adding partitions. Yet, sharding is not sufficient in and of itself. Many specialized databases have been built on top of sharding that are optimized for very specific workloads.
Gone are the days where one size ...
