⚡ no hosts were patched in the making of this site

Stop managing servers. Start shipping.

Serverless Society is a community for builders who use AWS serverless to hand off the undifferentiated heavy lifting (provisioning, patching, scaling, capacity planning) and put that time back into the product.

0 hoststo patch, scale, or babysit
Pay per useidle capacity costs (close to) nothing
6 pillarsour yardstick for every service
01 · The manifesto

Why we love serverless

Every hour spent on AMIs, auto-scaling groups and kernel patches is an hour not spent on customers. Serverless moves the host-management line so far down the shared-responsibility model that your team's surface area becomes code, config and data, and nothing else.

No host overhead

AWS owns the OS, runtime patching, fleet health and capacity. You own the business logic.

Scales with demand

From zero to a traffic spike and back without a single scaling policy to tune at 2 a.m.

Faster time to value

Managed integrations (events, queues, workflows) mean less glue code and shorter lead time per feature.

What you stop doing

  • Patching OS images and runtimes
  • Right-sizing instances and tuning auto-scaling
  • Paying for idle servers overnight
  • Building retry/queue plumbing by hand
  • Running cron boxes that nobody owns

What you start doing

  • Shipping features in days, not sprints
  • Modeling the business as events
  • Treating cost as a per-request metric
  • Letting managed services do the retries
  • Writing specs, then wiring services to match them
02 · Know your planes

Control plane vs data plane

AWS splits nearly every service into two halves. Knowing which half you're calling is the difference between an architecture that rides out a bad day and one that falls over with it.

Control plane CRUD on resources

The APIs that create, configure and delete things: functions, tables, queues, roles. Complex, stateful, and called rarely.

Think
"Set up the kitchen"
Called by
IaC (CDK, SAM, CloudFormation, Terraform), consoles, pipelines
Volume
Low: deploy time only
Design rule
Keep it out of your runtime & recovery path

Data plane the actual work

The APIs that do the job the resource exists for: invoke, read, write, send, receive. Simpler, so they're built for higher availability.

Think
"Cook the meals"
Called by
Your application code, every request
Volume
High: millions of calls
Design rule
This is what your hot path should depend on
ServiceControl plane (deploy time)Data plane (run time)
LambdaCreateFunction UpdateFunctionConfiguration PutProvisionedConcurrencyConfigInvoke
DynamoDBCreateTable UpdateTable CreateGlobalTableGetItem PutItem Query TransactWriteItems
S3CreateBucket PutBucketPolicy PutBucketLifecycleConfigurationGetObject PutObject
SQSCreateQueue SetQueueAttributesSendMessage ReceiveMessage DeleteMessage
EventBridgePutRule PutTargets CreateEventBusPutEvents
Step FunctionsCreateStateMachine UpdateStateMachineStartExecution SendTaskSuccess
IAM / STSCreateRole PutRolePolicyAssumeRole (authn on every call)
R1
Static stabilityPre-provision what you need to survive a failure. A recovery plan that calls CreateTable mid-incident is a plan that depends on the control plane.
R2
Never mutate infra in a requestNo CreateQueue inside a Lambda handler. Resources are born in IaC; handlers only use them.
R3
Separate the blast radiusPipelines own the control plane, workloads own the data plane. Different roles, different permissions, least privilege on both.
R4
Cache control-plane readsConfig, secrets and endpoints are fetched at init and cached (e.g., Lambda extensions / Powertools parameters), not on every invoke.
03 · The yardstick

Well-Architected pillars as our metrics

We grade every service and pattern against the six pillars of the AWS Well-Architected Framework, viewed through the Serverless Applications Lens. Pick a pillar to see the question we ask and the numbers we watch.

Service scorecard

Serverless Society editorial ratings (1–5): how much each service does for that pillar out of the box. Opinion, not an AWS benchmark.
04 · Superpowers

Patterns that feel like superpowers

Battle-tested AWS serverless patterns that turn a small team into a large one. Tap any card for the wiring, the plane split, and the pillar it boosts.

Francis Njoh, founder of Serverless Society, smiling outdoors in a navy blazer
05 · The founder

Great ideas deserve a fast path to production.

I'm Francis Njoh, a cloud builder and the founder of Serverless Society.

Most ideas don't die because they're bad. They stall somewhere between "this would be great" and "it's live", buried under provisioning, patching, scaling and babysitting servers. Every hour spent there is an hour not spent on the thing people actually use.

That's why I love serverless. Less infrastructure upkeep means more time creating. Hand the undifferentiated heavy lifting to managed services, and what's left for the team is the part that matters: code, config and data.

My goal is a software factory: an assembly line for ideas, where good patterns are captured once and reused everywhere. Serverless Society is where I'm building it in the open, starting with this site, which runs on the same patterns it writes about.

Shorter cyclesidea to production in days, not quarters
Repeatable deliverytemplates and pipelines, not heroics
More shippedtime moves from upkeep to building

Connect on LinkedIn →

06 · Join the society

Spend your time on the product, not the plumbing.

Monthly field notes: one pattern, one cost teardown, one spec-to-stack walkthrough. No servers were harmed.

One email a month. We store your address only to send the field notes. See exactly how it's stored →

The Serverless Pledge
  1. I will not SSH into production.
  2. I will keep the control plane out of my hot path.
  3. I will make every handler idempotent.
  4. I will let a managed service retry before I write a loop.
  5. I will measure cost per transaction, not per server.