Latency is a technical term used in Information Technology to express the time taken to transfer data from one IT components to another.
It often falls down in the networking part, but one has to keep in mind that latency is caused by both the physical part (cables type, distance, number of hops, etc.) and the software part (codec, multiplexing, protocol, volume, etc.)
Microservices architecture can improve latency:
- Decomposing/cutting by domain/functions is likely to reduce the amount of data per transaction (as opposed to macro data exchanges). For example, if you have Payment API and you need to initiate a payment by posting the Payment Details + Invoice + foreign currency rate in one request-response transaction, you could use the orchestration of 3 microservices with Payment Initiation API + Forex API + Invoicing API, with each of them requiring less amount of data transfer for each transaction “channel”.
- Scaling out the number of the same microservice is likely to keep the response time optimal, reducing the chance to add latency during the period of high load or peak load.
