Tradeoffs in distributed systems

Trade-offs in Distributed Systems
1. Compared to in-process function calls, remote calls are much slower. If multiple services need to be called in a transaction, then the latency increases significantly. If the volume of data to transfer is more, then it adds more latency to the transaction.
2. Because of network intervention, remote calls are more likely to fail, especially with a large number of services and communication, and the number of failure points increases.
3. Observability like full stack tracing is important because when an issues arises, developers responsible for different services find it difficult to identify the root cause.
4. Distributed transactions is another concern that do not have a perfect solution. We should build the system and tools for addressing consistency issues.
There are two solutions to address some of these issues
1. Reduce number of invocations by using batch processing. Instead of processing individual requests and invoking the service multiple times, we can consider to use batch invocation where ever it is possible.
2. Using asynchronous communication for services that do not have strong dependencies and do not need to be invoked directly.
#microservicesarchitecture #cloudnative

[@Shruti68-b9j]

Good to listen to you after a long time nagesh.
Looking forward for more transforming content