Which performance metric did the MIT study highlight when adding microservices—latency or throughput?

Interesting read! Did you find the author's argument on scalability convincing?

I've seen similar claims, but isn't the added complexity of microservices a major drawback for smaller teams or projects with limited resources?