Hands-On Microservices with Kubernetes

Storing your data in memory

The highest performance, as far as read and write latency and throughput is concerned, is when you keep your data in memory. There are different memory types and caches, but the bottom line is that memory is super fast. However, memory has significant downsides too, such as the following:

• A node has much more limited memory compared to disks (that is, it requires more machines to store the same amount of data).
• Memory is very expensive.
• Memory is ephemeral.

There are some use cases where you require your entire dataset in memory. In these cases, either the dataset is very small, or you can split it across multiple machines. If the data is important and can't be easily generated, then you can address the ephemeral ...