Rethinking Microservices: Why Google, Amazon and Others Are Moving Away from Traditional Microservice Architectures

Google Says Our Microservices Are Wrong!

In June 2023 a group of Google engineers led by Michael Whittaker published a paper titled “Towards Modern Development of Cloud Applications,” criticizing the current microservice paradigm for mixing logical and physical boundaries and causing performance, traceability, and management problems.

They propose a "microservice 2.0" approach that treats the application as a logical whole while delegating deployment decisions to an automated runtime, achieving up to 15× lower latency and 9× lower cost.

The three guiding principles are: (1) encourage developers to write single‑application logical components, (2) defer physical distribution and execution challenges to runtime, and (3) enable atomic deployment of the whole application.

Amazon Prime Video Team: Abandon Microservices, Adopt Monolith

In the same month, Amazon Prime Video released a case study showing that switching from a serverless, microservice‑based monitoring system to a monolithic architecture reduced operational costs by over 90% and simplified scaling.

The original solution relied heavily on AWS Step Functions, which hit concurrency limits and incurred high per‑state‑transition charges, prompting the team to rebuild the entire infrastructure as a monolith.

Others Also Giving Up Microservices

Companies such as Uber, Managed by Q, and many mid‑size teams have found that proliferating microservices leads to excessive monitoring, testing, CI/CD, and SLA overhead, especially when engineering resources are limited.

These teams are consolidating services back into larger monoliths or fewer, more manageable services, emphasizing the importance of choosing the right architecture for the right problem.

The Illusion of Microservice Prosperity: From Monolith to “Distributed Monolith”

Critiques highlight performance penalties from network serialization, difficulty tracing bugs in distributed systems, management complexity from independent release cycles, and fragile APIs that force unnecessary expansion.

Often, teams split a monolith into tightly coupled “microservices” without clear domain boundaries, effectively creating a distributed monolith that fails to deliver the promised benefits.

Google Proposes a New Kind of Microservice

Google’s paper separates programming models from deployment models, allowing developers to focus on code while the runtime optimizes cost‑effective execution, delivering microservice‑like advantages with better performance and lower cost.

A Year of Rethinking Infrastructure

Beyond microservices, cloud computing itself is being questioned; 37signals (Basecamp) moved away from public cloud to on‑premise servers, and FinOps practices are emerging to curb soaring cloud bills.

These trends suggest a broader industry reassessment of architectural choices made over the past decade.

Reference links: https://thenewstack.io/year-in-review-was-2023-a-turning-point-for-microservices/ https://dl.acm.org/doi/10.1145/3593856.3595909