REReact · Architecture
Community homeHot discussions Latest posts Best of Knowledge base Resources About

React · Architecture · react-design under load on an 8-core 16GB box — full benchmark scripts included

91
REr/react-design·posted by bob_chen·3 days agoPostmortem

react-design under load on an 8-core 16GB box — full benchmark scripts included

Short version: react-design needs almost no tuning at small and medium scale — the point where it starts to hurt is much further out than most people assume. Full measurements below.

We also fixed monitoring along the way: replaced average-based alerts with percentiles and split them per endpoint. False alerts dropped by about seventy percent and the on-call rotation visibly cheered up.

Image placeholder · object storage in production
POST /api/uploads → CDN origin pull
47 comments

47 comments

M
Wwinter·1 hour ago

Has anyone run a controlled experiment? I did, reducing it to a single variable, and the difference was 4% — within noise. So I suspect the main cause is something else.

495
Kkite·2 days ago

Worth learning from this debugging approach. We went straight at the logs and took a much longer route.

80
Hhuang_ke·2 days ago

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

240
Mmike_xu·5 hours ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

50
Cchen_dev·2 days ago

There is actually a simpler fix that needs no architecture change: move this check up to the gateway and the problem disappears. The cost is one extra lookup at the gateway.

13
Lli_ming·yesterday

Agreeing with the above. One addition: with this option enabled the GC count in your metrics doubles, so adjust the alert threshold at the same time or it will keep firing.

25
Ddev_zhouMod·2 days ago

Worth learning from this debugging approach. We went straight at the logs and took a much longer route.

451
Kkernel_panic·28 minutes ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

440
Ddev_zhouMod·2 days ago

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

409
Llinlin·12 minutes ago

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

340
Rran_bo·yesterday

Can you give a minimal reproduction? I ran it locally for ten minutes and could not reproduce on macOS with the latest version.

340
Hhuang_keOP·2 days ago

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

1
Zzhu_zong·just now

This is not a react-design problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

201
Zzhu_zongOP·28 minutes ago

One counter-example: below react-design 7.4 the semantics of that code are different, so do not copy it verbatim. We got burned in staging and rolled back once.

131
Ddev_zhou·28 minutes ago

Saved. I am reworking this area this week — this saves a lot of wrong turns.

396
Llinlin·2 days agoedited

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

17
Nnikic·28 minutes ago

One counter-example: below react-design 7.4 the semantics of that code are different, so do not copy it verbatim. We got burned in staging and rolled back once.

272
WwinterMod·5 hours agoedited

Agreeing with the above. One addition: with this option enabled the GC count in your metrics doubles, so adjust the alert threshold at the same time or it will keep firing.

144
Kkernel_panic·2 days ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

6
Hhuang_ke·2 days ago

Agreeing with the above. One addition: with this option enabled the GC count in your metrics doubles, so adjust the alert threshold at the same time or it will keep firing.

184
Rran_bo·2 hours ago

There is actually a simpler fix that needs no architecture change: move this check up to the gateway and the problem disappears. The cost is one extra lookup at the gateway.

1
Kkite·1 hour ago

Saved. I am reworking this area this week — this saves a lot of wrong turns.

178
Kkite·28 minutes ago

Worth learning from this debugging approach. We went straight at the logs and took a much longer route.

144
KkiteMod·2 hours ago

This is not a react-design problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

447
Ttang_hao·2 days ago

This is not a react-design problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

133
Llinlin·12 minutes ago

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

92
Kkite·3 minutes ago

Can you give a minimal reproduction? I ran it locally for ten minutes and could not reproduce on macOS with the latest version.

48
Cchen_dev·3 minutes ago

I just read the react-design source — the author actually explains the reasoning in a comment, roughly "so that it degrades into predictable behaviour in extreme cases".

43
Oops_wang·yesterday

Sharing our numbers, 8 cores 16GB, same scenario:

| Concurrency | P50 | P99 |
|---|---|---|
| 200 | 12ms | 88ms |
| 500 | 31ms | 340ms |

P99 clearly collapses at 500 concurrency, which lines up with your knee point.

333
Zzhu_zongOPMod·28 minutes agoedited

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

421
Sswoole_leeOP·28 minutes ago

Has anyone run a controlled experiment? I did, reducing it to a single variable, and the difference was 4% — within noise. So I suspect the main cause is something else.

434
Ttang_hao·2 days ago

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

159
Lli_mingMod·28 minutes ago

One counter-example: below react-design 7.4 the semantics of that code are different, so do not copy it verbatim. We got burned in staging and rolled back once.

12
RraseOP·2 days ago

Sharing our numbers, 8 cores 16GB, same scenario:

| Concurrency | P50 | P99 |
|---|---|---|
| 200 | 12ms | 88ms |
| 500 | 31ms | 340ms |

P99 clearly collapses at 500 concurrency, which lines up with your knee point.

138
Rran_bo·2 days ago

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

75
Hhuang_ke·2 days agoLevel 6

Sharing our numbers, 8 cores 16GB, same scenario:

| Concurrency | P50 | P99 |
|---|---|---|
| 200 | 12ms | 88ms |
| 500 | 31ms | 340ms |

P99 clearly collapses at 500 concurrency, which lines up with your knee point.

5
Mmike_xu·2 hours ago

I just read the react-design source — the author actually explains the reasoning in a comment, roughly "so that it degrades into predictable behaviour in extreme cases".

3
Nnikic·5 hours ago

Saved. I am reworking this area this week — this saves a lot of wrong turns.

77
Wwinter·just now

Has anyone run a controlled experiment? I did, reducing it to a single variable, and the difference was 4% — within noise. So I suspect the main cause is something else.

368
Zzhou_yi·2 hours ago

Saved. I am reworking this area this week — this saves a lot of wrong turns.

345
Cchen_devMod·2 days agoedited

There is actually a simpler fix that needs no architecture change: move this check up to the gateway and the problem disappears. The cost is one extra lookup at the gateway.

273
Bbob_chen·just now

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

26
Kkernel_panic·just now

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

441
Cchen_dev·2 days ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

6
Lli_mingOP·3 minutes ago

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

90
Oops_wang·2 days ago

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

3
Ttang_hao·5 hours ago

Can you give a minimal reproduction? I ran it locally for ten minutes and could not reproduce on macOS with the latest version.

2

This is the post detail page /en/c/react-design/post/p3. Posts and comments are generated deterministically from a seeded PRNG, so the same post always renders the same content and the link can be shared, reloaded and indexed. In production this page reads MySQL for the post, Redis for hot-post caching, and fetches the whole comment tree in a single query on the path column.

See the database schema →