Work

Open SourceGitHub, npm, PyPI, and public packages.

Depth

JourneyCareer timeline, roles, and education.AchievementsScholarships, rankings, and milestones.

Writing

PublicationsResearch-facing work and technical reports.ProfilesAll key public links in one place.
When to Use Serverless, Containers, or Both
Cloud

When to Use Serverless, Containers, or Both

Serverless versus containers isn't really a technology debate — it's a workload-shape question. Here's how to make the call based on what the team can actually sustain.

Published December 1, 20259 min readUpdated Aug 31, 2026

Written by · Full-Stack Agentic AI Software Engineer — AI Agents, Automation & Revenue Systems for GTM/RevOps teams

Related paths

Cloud ServiceAppNaviOpen SourceContact

In brief

When should you choose serverless, containers, or both?

Serverless wins for bursty, event-driven workloads where teams value fast iteration and lower overhead. Containers win when runtime control, long-lived processes, or custom execution environments matter. Hybrid usually wins when event ingestion and lightweight APIs stay serverless while steady workloads live in containers.

  • Serverless wins for event-driven, bursty workloads with fast iteration needs.
  • Containers win when runtime control and operational maturity are core to the product.
  • Hybrid is often the real answer, mixing both approaches for different workload shapes.

Evidence notes

Project evidence

The AppNavi Observability Platform and Legacy Monorepo projects show these patterns in practice.

Evidence boundary

This article frames infrastructure choices around workload and team maturity. It does not claim one pattern is universally superior.

The serverless-vs-containers debate is really about workload shape

Nobody actually needs to choose serverless or containers in the abstract. The real question is narrower: what does this workload look like, how often does its shape change, and how much operational ownership can the team actually sustain — not aspire to, sustain.

Anchor the debate to a real system and it gets easier fast. The AppNavi Observability Platform and the Legacy Monorepo and Microservice Modernization project are both good examples of teams making this call under real constraints instead of in a vacuum.

Serverless usually wins when

  • The workload is event-driven or genuinely bursty, not just spiky-looking on a chart.

  • The team is optimizing for iteration speed over infrastructure control.

  • Execution boundaries are naturally small and useful on their own.

  • This pairs well with the Cloud Architecture service and most data or notification workflows.

Containers usually win when

  • Runtime control matters more than elastic convenience.

  • Long-lived processes or a custom execution environment sit at the center of the product.

  • The team already has the operational maturity to run infrastructure and wants the predictability that comes with it.

Hybrid is often the real answer

Most teams I've seen end up here anyway: event ingestion, automation, and thin APIs stay serverless, while steady-state workloads and anything stateful move into containers. It isn't indecision. It's matching the tool to the shape of the job twice instead of once.

Pair this with From 300M Events to Usable Insight for the cost side, or How to Modernize a Legacy Monorepo Without Freezing Delivery for the delivery side.

The real decision criteria

Infrastructure should serve the workload, not a team's identity as a serverless shop or a Kubernetes shop. The strongest platform decisions match execution shape, cost profile, and operating maturity all at once. If you need help making that call, reach out.

Apply this article

How to turn insights into execution

A practical sequence for teams turning concepts into production outcomes.

ServerlessContainersCloud ArchitectureInfrastructurecloud

Audit your current state

Map the bottlenecks and constraints connected to the article’s core problem.

Choose one bounded change

Test the most useful recommendation on one workflow before widening the scope.

Measure what changed

Keep the parts that improve the work, document what failed, and make the next decision from evidence.

Next Steps

Continue reading

VPS vs PaaS vs Serverless: What Should a Startup Actually Run On?

Pick based on the shape of your cost curve, not on what's trendy this month. A VPS, a managed PaaS, and serverless don't compete on price — they compete on where the bend in that curve sits.

Navigate

Kubernetes vs. Serverless for AI Workloads: A Practical Comparison

GPU scheduling, cold starts, and autoscaling break serverless assumptions. Learn when Kubernetes is mandatory for AI—and when it's overkill.

Navigate

AWS Serverless Cost Optimization: A Real Audit Walkthrough

Serverless teams overspend by 30-50%. Here's how to audit your Lambda, API Gateway, and NAT costs, find the leaks, and cut your bill.

Navigate

Case Study: 300M Events and a 12x Query Performance Gain

How disciplined diagnosis and architectural optimization cut query latency by 12x on a 300M event/day analytics platform.

Navigate

Vercel Pricing Explained: Where the Bill Actually Comes From

Vercel publishes every rate it charges. The problem is there are nine meters, two of them have a real free tier, and seven don't. Here's the anatomy, straight from the pricing page.

Navigate

The Hidden Costs of Self-Hosting: A Realistic Monthly Bill

The invoice is the easy half, and it's small. Line it up from published rates, then look at the half no invoice tracks — the migration weekend, the patching, the on-call, the things you now own that used to be someone else's problem.

Navigate

Who Controls Your DNS, and What Is That Control Worth?

One DNS provider's bad Friday in October 2016 took Twitter, Netflix, Spotify and Reddit offline at once — not because any of them failed, but because they'd all quietly delegated the same single switch to the same single company.

Navigate

What's the Difference Between Renting and Owning Your Infrastructure?

A domain is a lease you renew on your own terms. A social account, a SaaS subscription, a hosted storefront is a lease the landlord can end on theirs. The difference was never about cost — it's about who holds the option to say no.

Navigate

Cloud Cost Review Checklist for Founders Who Don't Have a Platform Team Yet

Your cloud bill doesn't have to be a surprise. A practical checklist to find waste, set guardrails, and understand what you're actually paying for.

Navigate

CI/CD Pipeline Design for Fast-Moving Product Teams

Trunk-based development, feature flags, and progressive deployment strategies enable teams to ship faster while managing risk. Here's how to structure a CI/CD pipeline that scales with velocity.

Navigate

Where to go next

Filter the archive

Beyond the archive