2023: The Year Microservices Were Questioned – Google, Amazon and Others Shift to Monoliths

2023 has been described as the "year of microservices reversal," with major players such as Google, Amazon, Uber, and others re‑evaluating the long‑standing belief that microservices are the default cloud‑native architecture.

Google engineers, led by Michael Whittaker, published the paper Towards Modern Development of Cloud Applications , arguing that traditional microservices mix logical and physical boundaries, causing performance and operational penalties. Their proposed "microservices 2.0" builds applications as logical monoliths that are handed to an automated runtime, achieving up to 15× lower latency and 9× lower cost.

Amazon Prime Video released a case study showing that abandoning serverless and microservices for a monolithic architecture reduced operational costs by over 90% and simplified scaling, turning a previously costly Step Functions‑based pipeline into a more efficient single‑app solution.

Other companies such as Uber and Managed by Q have reported similar experiences: the overhead of maintaining dozens of microservices in small teams leads to monitoring, testing, CI/CD, and SLA challenges, prompting them to consolidate services into larger, more manageable units.

The article lists common microservices drawbacks: network‑induced latency, difficulty tracing bugs, management complexity from independent release schedules, and fragile APIs that hinder evolution.

Google’s new principles encourage developers to write applications as logical single‑components, defer physical deployment decisions to the runtime, and aim for atomic deployments, promising higher performance and lower cost.

While some argue that microservices are still appropriate at very large scale, Google’s separation of programming and deployment models offers a middle ground that retains microservice benefits without the associated penalties.

The broader infrastructure rethink also touches on cloud computing itself, with figures like DHH (David Heinemeier Hansson) abandoning cloud services in favor of on‑premise hardware, and the rise of FinOps to control soaring cloud bills.

Finally, the author adds a personal note encouraging readers to like, share, and support the content, and promotes a paid knowledge community offering deep‑dive courses on Spring, Kafka, K8s, and other enterprise technologies.