I'm unhappy with paying for always-on servers, should I switch to serverless?

Maybe — but only if your workload fits event-driven, bursty, or intermittent traffic. Serverless can reduce cost and ops burden, but it’s not automatically cheaper or simpler.

Good reasons to switch

Serverless is usually a strong fit if:

• Traffic is spiky or mostly idle
• You want to avoid managing servers
• Your app is event-driven:
  • HTTP APIs
  • background jobs
  • webhooks
  • scheduled tasks
  • queue consumers
• You can tolerate:
  • cold starts
  • some platform limits
  • more complex debugging/observability

Bad reasons / warning signs

Serverless may not be ideal if you have:

• Steady high traffic all day
• Long-running processes
• Low-latency requirements
• Stateful workloads
• Heavy use of:
  • in-memory caching
  • persistent connections
  • custom runtimes / native dependencies
• Costs driven by:
  • lots of invocations
  • high outbound network traffic
  • expensive per-request startup work

In those cases, always-on containers/VMs may actually be cheaper and more predictable.

Cost reality

Serverless pricing is often:

• cheaper when usage is low or uneven
• more expensive when:
  • requests are frequent
  • functions run long
  • you need high concurrency
  • you pay for lots of cold starts or orchestration

A common mistake is comparing:

• VM monthly bill vs
• serverless compute alone

You also need to include:

• database costs
• logging/monitoring
• networking
• API gateway
• queues/events
• storage
• retries/timeouts

A practical middle ground

Before going all-in, consider:

• autoscaling containers on ECS/Kubernetes/Cloud Run/App Runner
• smaller always-on instances
• scheduled scale-down
• spot/preemptible instances
• serverless for edge cases, not everything

Rule of thumb

Switch to serverless if:

• workload is irregular,
• you want less ops,
• and latency/cost predictability are not top priorities.

Stay with servers/containers if:

• workload is steady,
• performance matters,
• or the app is already simple and cheap to run.

Best next step

If you want, I can help you decide by estimating cost tradeoffs for your specific app.
Send me:

• current traffic pattern
• stack/cloud
• average request duration
• whether it’s API, jobs, or both
• current monthly server cost

and I’ll tell you whether serverless is likely to save money.