Serverless Scalability: Myths, Quotas & Smart Architecture
Vadym Kazulkin kicked off with a reality check: AWS Serverless services don’t scale without limits.
- Key quotas for API Gateway, Lambda, DynamoDB, Aurora Serverless, SQS, S3 and more
- Architectural choices between DynamoDB vs. Aurora
- Event-driven design with SQS, SNS, Kinesis, EventBridge
- Practical concepts like token bucket algorithms, concurrency handling, and exponential backoff
View Vadym’s Full Session Details
Zero-Cost Serverless? Yes, Really.
Dan Erez showed how you can use your existing infrastructure — Kubernetes, open-source serverless platforms, and even browsers — to run workloads for free, at scale.
- How to spot underused resources in your org
- The open-source stack to make them serverless-ready
- When it’s worth running in-house vs. public cloud
View Dan’s Full Session Details
📅 Join Us in Berlin – October 20–23, 2025
Until September 18th, you can save up to €285 on your tickets!
FREQUENTLY ASKED QUESTIONS
What are the scalability limitations of AWS Serverless services?
AWS Serverless services do not scale infinitely. Service-specific quotas apply to Lambda, API Gateway, DynamoDB, Aurora Serverless, and SQS, and must be accounted for in architecture design.
How does AWS handle concurrency in Serverless applications?
Concurrency in AWS Lambda is managed through regional and function-level limits. Patterns like token bucket algorithms and exponential backoff help mitigate overload scenarios.
What are the architectural trade-offs between DynamoDB and Aurora Serverless?
DynamoDB offers low-latency and high scalability, while Aurora Serverless provides relational features and consistency. The choice depends on workload requirements.
Which AWS services are key in event-driven architectures?
AWS services like SQS, SNS, Kinesis, and EventBridge are essential for event-driven design, enabling decoupled, scalable microservices communication.
Can you really run serverless workloads at zero cost?
Yes. By using existing infrastructure like Kubernetes clusters or browser runtimes and open-source tools, zero-cost serverless workloads are achievable in some cases.
