In recent years, the adoption of serverless architecture has grown significantly in enterprise environments. However, the decision to migrate an established infrastructure to this model still raises doubts. On one hand, people talk about greater agility, usage-based billing, and reduced operational overhead. On the other hand, there are concerns about latency control, debugging complexity, and vendor lock-in. What criteria do you consider most critical when evaluating a transition to serverless? Have you encountered cases where the benefits clearly outweigh the risks, or do you prefer a hybrid approach? I’d love to hear the community’s experiences and opinions.
Should we migrate our infrastructure to serverless cloud architecture?
👁️ 130 views💬 1 replies❤️ 0 likes
1 Replies
In our latest project, we migrated a home sensor management API to AWS Lambda and DynamoDB. The first rule we followed was measuring the **load profile**: if most invocations are sporadic or unpredictable, serverless only pays per execution and reduces costs; if the load is constant and high, the "pay-per-use" model can end up being more expensive than reserved instances. Another key criterion is **latency**: we tested the function in a pre-production environment and measured the "cold start." When the startup time exceeded 200 ms, we opted to enable provisioned concurrency to ensure consistent times, which added a fixed cost but maintained an acceptable user experience.
Regarding **debugging and observability**, I integrated CloudWatch and X-Ray from day one and set up structured logs sent to Elasticsearch; without that visibility, debugging in serverless can become a headache. Finally, to avoid **vendor lock-in**, I implemented an abstraction layer using our API's OpenAPI SDK so that, if we ever want to move to Google Cloud Functions or Azure, the change is reduced to a few configurations. In practice, combining serverless for high-variability points with containers or VMs for critical and constant-latency components gave us the best of both worlds: agility and control.