Evolution of business logic from monoliths through microservices, to functions
adrian cockcroft

Awesome Post!

LSM-tree storage model might not be a great fit for SSDs for all kinds of workloads because (a) the difference between random and sequential performance is not as large as with HDDs. An LSM-tree that performs a large number of sequential I/Os to reduce later random I/Os may be wasting bandwidth; (b) SSDs have a large degree of parallelism, but not all LSM designs benefit from that; (c) LSM-trees have high write amplification hence SSDs can wear out through repeated writes. In fact, LSM-based technology is tied closely to its usage upon classic HDDs.

I agree though that they are cost effective, highly available and scalable databases and hence a great fit a microservice deployment.