Small now, sharp later: API versioning without regrets
Additive changes, date-based versions, and deprecations users can plan around.
The numbers below are from my own machine and my own workload. Treat them as a method, not a benchmark: the point is how the measurement was taken, not the exact figures it produced.
Where I started
Tools matter less than the feedback loop they enable. A fast, boring loop beat every sophisticated setup I tried, and I've kept it ever since, even as the rest of the stack changed twice over.
There is a version of this article that opens with theory. I decided against it, because the theory is easy to find and the motivation is not — so let's start with the thing that actually broke.
What went wrong
The numbers below are from my own machine and my own workload. Treat them as a method, not a benchmark: the point is how the measurement was taken, not the exact figures it produced.
I want to be honest about the trade-offs here. The approach I settled on solves the problem I had, not every problem of this shape, and there is a clear point where you should reach for something more heavy-duty.
A note for context that helps the reader.
[deploy]
strategy = "rolling"
healthcheck = "/healthz"
max_unavailable = 1The fix
The numbers below are from my own machine and my own workload. Treat them as a method, not a benchmark: the point is how the measurement was taken, not the exact figures it produced.
I want to be honest about the trade-offs here. The approach I settled on solves the problem I had, not every problem of this shape, and there is a clear point where you should reach for something more heavy-duty.
The best engineering advice I've received is unglamorous: make it measurable, make it reversible, and only then make it fast.
What I'd do differently
The first version worked, which is exactly why it was hard to see how wrong it was. Everything looked fine in the demo, and everything looked fine until traffic arrived on a Tuesday afternoon and the graph started doing something I'd never seen before.
There is a version of this article that opens with theory. I decided against it, because the theory is easy to find and the motivation is not — so let's start with the thing that actually broke.
| Approach | Best for | Watch out for |
|---|---|---|
| Baseline | Quick comparisons | No tail visibility |
| Instrumented | Finding the real hotspot | Overhead of the probes |
| Production | Truthful numbers | Harder to reproduce |
Where I started
Tools matter less than the feedback loop they enable. A fast, boring loop beat every sophisticated setup I tried, and I've kept it ever since, even as the rest of the stack changed twice over.
The numbers below are from my own machine and my own workload. Treat them as a method, not a benchmark: the point is how the measurement was taken, not the exact figures it produced.
A note for context that helps the reader.
What went wrong
There is a version of this article that opens with theory. I decided against it, because the theory is easy to find and the motivation is not — so let's start with the thing that actually broke.
What surprised me most was how small the final change was. Most of the work was understanding the problem precisely enough that the fix became obvious, and that understanding took far longer than the fix itself.
CREATE INDEX CONCURRENTLY idx_orders_created_at
ON orders (created_at DESC)
WHERE status = 'open';The fix
I want to be honest about the trade-offs here. The approach I settled on solves the problem I had, not every problem of this shape, and there is a clear point where you should reach for something more heavy-duty.
There is a version of this article that opens with theory. I decided against it, because the theory is easy to find and the motivation is not — so let's start with the thing that actually broke.
- Latency at p50, because most users live there
- Latency at p99, because someone always lives there
- Error rate, because a fast failure is still a failure
- Saturation, because averages hide the cliff
What I'd do differently
The first version worked, which is exactly why it was hard to see how wrong it was. Everything looked fine in the demo, and everything looked fine until traffic arrived on a Tuesday afternoon and the graph started doing something I'd never seen before.
There is a version of this article that opens with theory. I decided against it, because the theory is easy to find and the motivation is not — so let's start with the thing that actually broke.
By the end of this post you should be able to reproduce the whole thing yourself in under an hour, including the parts I got wrong the first time — those are marked so you can skip them if you're in a hurry.
Wrap-up
If you remember one thing, make it this: additive changes, date-based versions, and deprecations users can plan around compounds. The cost of getting it right early is a afternoon; the cost of getting it late is a migration.