Video summary

AWS re:Invent 2024 - Building an integration strategy for the future (API208)

Main summary

Key takeaways

Technology

Summary

The session argues that integration should be treated as a long-term business strategy—not just a series of one-off connections. A good strategy can help organizations move faster, reduce risk, and adapt their systems incrementally, even when they have many legacy applications and varied technology stacks.

Integration strategy principles

  • Start from business outcomes. Executives prioritize speed while also needing security, compliance, and appropriate guardrails. Developers can support that goal by experimenting with small, reversible changes.
  • Think big, start small, scale fast. Test ideas in low-risk ways, then expand what works. Cloud services can make experimentation more reversible.
  • Plan for organizational change as well as technical change. Event-driven architectures, for example, may require teams to align around business domains and own their services and infrastructure.
  • Avoid creating tomorrow’s legacy systems. A strategy helps teams make consistent decisions across future projects, rather than repeatedly choosing integrations from scratch.

The speakers compare four broad approaches:

  1. No defined strategy: Can suit an early-stage organization or allow room for experimentation, but may lead to inconsistent decisions and missed opportunities.
  2. HTTP/API for everything: Simple and widely supported, but can create tightly connected “spaghetti” systems and may be less suitable for asynchronous, long-running processes or some data workflows.
  3. Enterprise service bus (ESB): Offers comprehensive, packaged integration capabilities and can work well for centralized teams, but may involve high upfront costs and create organizational or technical bottlenecks.
  4. Cloud integration strategy: Uses a range of services to fit different needs. The session groups these needs into application integration (including APIs and event-driven architectures), data integration, and file integration.

Customer examples

KnowBe4

KnowBe4’s overnight batch processing grew slower as its data and customer base expanded, sometimes leaving customer-facing data stale. The company moved toward near-real-time, event-driven processing:

  • Teams published events through Amazon EventBridge, with teams organized around business domains and services in separate AWS accounts.
  • A common processing pattern used Amazon SQS to buffer events, AWS Lambda to process them, and Aurora Serverless v2 for team-managed data.
  • AWS AppSync and GraphQL provided a unified way to retrieve data from multiple databases.
  • The speakers said data delivery went from hours to seconds. The project took four months and produced 235 business events. The subtitles give conflicting figures for total messages, so that number is unclear. They also report 99.99% availability and that subsequent workloads were delivered a quarter ahead of schedule.

Regeneron

The biotech company needed to keep supporting SFTP workflows used by internal and external parties while reducing the burden of scaling, securing, and maintaining its file-transfer system.

  • It adopted AWS Transfer Family as a managed replacement that preserves familiar file-transfer protocols and stores files in Amazon S3.
  • S3 lifecycle policies moved infrequently accessed data to Glacier storage. AWS KMS supported encryption, while API Gateway and Lambda were used for custom authentication and authorization.
  • The project took four months. The speakers report an 80% reduction in storage costs and a 90% reduction in overall solution costs, along with lower infrastructure-management effort.

NASA Jet Propulsion Laboratory (JPL)

For a public “send your name to Mars” campaign, JPL faced unpredictable demand. A serverless design using AWS Lambda handled growth that could not have been reliably forecast: the team expected two million names over 90 days but reached that volume in two hours.

JPL also used a human-in-the-loop approach to train an algorithm to filter inappropriate submitted names, freeing engineers from manual review.

Nationwide Children’s Hospital and a genomics research institute

The organizations used an integration-first, event-driven approach to connect physical processes, data analysis, and cloud services for clinical genomics workflows. Event triggers and AWS Step Functions enabled new workflows to be added without rewriting existing ones.

A process that previously took weeks was reduced to about a day, helping clinicians get relevant information sooner.

Guides and further learning

This was presented as a 200-level strategic session, rather than a detailed service tutorial. The speakers pointed attendees to a QR-linked collection of resources, including customer case studies and material on event-driven architecture. They also referenced two blog posts covering the KnowBe4 example and a blog post on Regeneron. Additional technical sessions were recommended for deeper service-level detail.

Main speakers and sources

  • Justin — AWS director responsible for Amazon EventBridge and AWS Step Functions; discussed integration strategy and cloud integration patterns.
  • Tom Soderstrom — AWS Enterprise Strategist and former NASA/JPL technology executive; shared executive priorities and JPL experience.
  • Brian — AWS serverless specialist Solutions Architect; presented the customer case studies. His surname is unclear in the subtitles.
  • Source: AWS re:Invent 2024 session, “Building an integration strategy for the future (API208).”

Rate this summary

Your feedback will help improve summaries.

Improve this summary

Reprocess with a stronger model when the summary feels incomplete or inaccurate.

Pro

Translate summary in another language

Pro

Ask questions to this video

Chat for follow-up questions, clarifications, and source-backed answers.

Coming soon

Share this summary

Original video