Back in the 2000s, SOAP/WSDL with ESB (Enterprise Service Bus) was the dominant server-side architecture for many companies. Since the 2010s, microservices and service mesh technologies have grown wildly and thus became the de-facto industry standards. By Xinyu Liu.
At Lyft, hundreds of microservices are backed by numerous databases. New tables are being created constantly to support features and demands of our fast-growing business. Online
The main sections of the article:
- A use case
- CDC pipeline remake
- Architecture
- Load balancing and auto scaling
- A greedy algorithm
- Threading model
During performance testing, the CDC adapter service routed 4 million messages per minute with an average message size of 1.5KB. Such a level of workload was easily handled by 5 Kubernetes vCPU and 10 GB memory in total.
Plenty of other resources are mentioned. And you will get charts explaining architecture etc. Good read!
[Read More]