The Taxi You Only Pay for When You Ride
why serverless data architecture exists — paying for compute only while it's actually doing work, not for the infrastructure sitting idle.
Part 1
Part 2
Before Everyone Owned Their Own Fleet
how organizations managed data infrastructure before serverless computing, and the operational burden that came with owning it all.
Part 3
No Driver Idling While You're Not Riding
how scale-to-zero, the ability for serverless infrastructure to shrink to nothing between jobs, eliminates idle cost entirely.
Part 4
The Dispatcher That Finds You a Car in Seconds
how automatic provisioning behind the scenes makes serverless infrastructure feel instantly available despite scaling from zero.
Part 5
The Wait Before the First Car Arrives
why cold starts — the brief delay when serverless infrastructure spins up from zero — matter, and how to design around them.
Part 6
Different Cars for Different Kinds of Trips
the different flavors of serverless data infrastructure — functions, containers, and databases — and how they fit different needs.
Part 7
The Meter Starts the Moment You Get In
how serverless billing granularity — down to the invocation, millisecond, or byte scanned — differs fundamentally from provisioned infrastructure pricing.
Part 8
No Garage of Your Own to Maintain
how serverless architecture removes patching, scaling, and capacity management entirely from the user's responsibility.
Part 9
When a Car Breaks Down, Another One Appears
how serverless infrastructure's built-in redundancy handles failures automatically, without the user needing to design for it explicitly.
Part 10
Sharing the Road With Every Other Rider
how serverless infrastructure's multi-tenancy works, and what it means for isolation, security, and occasional performance variability.
Part 11
Every Trip Has a Maximum Length
why serverless functions impose maximum execution time limits, and what that means for designing workloads that fit within them.
Part 12
Where Your Luggage Goes Between Trips
why serverless functions are stateless, and why persistent data has to live somewhere external, not in the function itself.
Part 13
A Car Shows Up Because Something Happened
how event-driven triggers — new files, messages, or database changes — are what actually kick off serverless data workloads.
Part 14
Querying Data Without Renting the Whole Garage
how serverless query engines and data warehouses let organizations analyze data without provisioning and managing dedicated infrastructure.
Part 15
A Pipeline Built Out of Individual Rides
how serverless ETL and streaming pipelines chain together individual event-driven steps into a coordinated, sustained data flow.
Part 16
Stringing Many Short Trips Into One Journey
how serverless orchestration tools coordinate multi-step workflows reliably, handling sequencing, retries, and error paths.
Part 17
When Owning the Car Actually Makes More Sense
the genuine limitations of serverless architecture, and the workload characteristics that still favor dedicated, provisioned infrastructure.
Part 18
The Fare That Adds Up on a Daily Commute
how to actually calculate the cost crossover point between serverless and provisioned infrastructure for a real workload.
Part 19
Tracking Every Car Across the Whole Network
why observability is genuinely harder in serverless architecture, and how distributed tracing helps make sense of it.
Part 20
The Whole Rideshare Network, Working Together
reassembling every piece covered across this series into the complete picture of what disciplined serverless data architecture looks like.