Capacity forecaster
A small system that measures real capacity, forecasts demand against it, and answers the only question that matters: will we be short, and is there still time to fix it? A forecast with a shorter horizon than your hiring lead time is decoration, and the post on that is the important one. Seven posts on the same system, one diagram at a time, with a cost breakdown and an engineering reference at the end.
- 01
A capacity forecaster on AWS for a few dollars a month
The whole system on one page — capacity, demand, gap — and the lead time that determines whether any of it is useful.
- 02
How the real capacity gets measured
Why capacity must be measured from output, what accounts for the gap between paper and actual, how the bottleneck is identified, and why it moves.
- 03
Why the peak matters more than the average
Why average utilisation hides shortfalls, how a range is produced instead of a point, why seasonality usually dominates, and how the peak is expressed.
- 04
How far ahead is worth forecasting
Why lead time sets the minimum horizon, the different lead times for different responses, where a forecast stops being informative, and reporting the notice period alongside the shortfall.
- 05
How a forecast gets scored afterwards
Why forecasts must be stored as they were made, how they are scored against outcomes, the systematic biases that emerge, and what to do about them.
- 06
What the capacity forecaster costs
About $1 a month. One Bedrock read per forecast run is the only line that grows; the queue, the table, the mail and the storage are rounding errors. Plus the three ways the bill could surprise you.
- 07
Engineering reference: the capacity forecaster architecture
Same system, drawn purely for engineers. Service names, region, Lambda inventory, IAM scopes, the schemas and the exact model id.
Frequently asked questions
- What is a capacity forecaster?
- A small serverless system that measures actual available capacity, projects demand against it, and flags where a shortfall is coming while there is still time to act.
- Why is paper capacity wrong?
- Because it assumes full attendance, no breakdowns, no setup time and no rework. Real available capacity is typically a good deal lower and the gap is measurable.
- Why forecast the peak instead of the average?
- Because you do not run out of capacity on an average week. The average tells you whether the business is viable; the peak tells you when you will fail to deliver.
- How far ahead should a forecast run?
- At least as far as the lead time to add capacity. A four-week forecast in a business where hiring takes twelve weeks cannot change any decision.
- What does it cost to run?
- A couple of dollars a month. See part six.