Key Takeaways from AWS re:Invent 2016: New Services, Data Architecture, and Operational Insights

New Product Highlights

Around 1,000 new AWS features were announced, spanning five core areas: compute, storage, networking, security, and tools. Highlights include custom FPGA instances, the Lightsail starter service, AWS Glue, AWS Batch, Athena (Presto‑based query), Lambda@Edge, Step Functions, and a suite of AI services.

1. Compute

A deep dive into IaaS updates such as custom hardware, FPGA instances, and the user‑friendly Lightsail. In the PaaS layer, services like AWS Glue, AWS Batch, Athena, and Lambda were emphasized, along with AI offerings that lower entry barriers for natural‑language, speech, and image recognition.

2. Storage

AWS introduced upgrades to Snowball and the massive data‑transport solution Snowmobile, illustrating the importance of efficient large‑scale data movement.

3. Networking

James Hamilton announced AWS’s own undersea fiber cables, underscoring the strategic need for high‑performance, private connectivity across AWS data centers.

4. Security

AWS Shield was launched to mitigate DDoS attacks, with expectations of further security services to strengthen overall protection.

5. Tools

The focus on API‑driven services continues, with X‑Ray highlighted as a unified APM solution, and additional DevOps tools reinforcing the vision that developers write application code while AWS handles the underlying infrastructure.

Theoretical Insights

Herd Era

CTO Werner Vogels introduced the slogan “There Are No Cattle, There Is Only The Herd,” extending the Pets‑vs‑Cattle analogy to emphasize serverless, where developers focus solely on business logic and let managed services handle the rest.

AWS Well‑Architected Framework

The updated framework outlines five pillars—Security, Reliability, Performance Efficiency, Cost Optimization, and Operational Excellence—highlighting trade‑offs and the importance of not compromising security or operational excellence.

Modern Data Architecture

English‑Fluent‑Talk’s “Three‑Step Data” model (Data in, Data processing, Data out) is presented, with a 数据三步走 code tag preserved. The model covers data ingestion (e.g., Flume, DataX), storage/processing choices (HDFS, S3, Kafka, Hive, Spark, Presto), and downstream usage (BI or product features). AWS’s own ten‑point Modern Data Architecture expands this lifecycle with corresponding services.

Vendor and Case Studies

Numerous AWS ecosystem partners exhibited, and Netflix shared extensive sessions on architecture, data, networking, and culture, including their regional disaster‑recovery drills (Chaos Monkey‑style).

Conclusion

Since 2014, English‑Fluent‑Talk has relied on AWS for backend services and processing over 360 million minutes of user voice data. Insights from re:Invent 2016 will guide future product development to better serve its 36 million users.