Hands-On Microservices with Kubernetes

Performance and cost

Performance and cost have a very interesting relationship. There are many ways to improve performance. Some of them reduce costs, like optimizing your code, compressing the data you send, or pushing computation to the client. However, other ways to improve performance increase the cost, like running on stronger hardware, replicating your data to multiple locations close to your client, and pre-fetching unrequested data.

In the end, this is a business decision. Even win-win performance improvements are not always like improving your algorithms is, at the high-priority. For example, you can invest a lot of time in coming up with an algorithm that runs 10x faster than the previous algorithm. But the computation time might ...