ECEdge Computing · Showcase
Community homeHot discussions Latest posts Best of Knowledge base Resources About

Edge Computing · Showcase · Discussion: is the edge-computing-show ecosystem being replaced by another stack?

364
ECr/edge-computing-show·posted by zhou_yi·32 minutes agoExperience

Discussion: is the edge-computing-show ecosystem being replaced by another stack?

Short version: edge-computing-show 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.

One last trap: in container environments remember to adjust the memory-related parameters in step. Otherwise the host limit and the process expectation disagree, and the symptom is intermittent, unreproducible failure.

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

69 comments

M
RraseOP·2 days ago

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

432
Bbob_chen·2 days ago

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

493
Ddev_zhou·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.

60
Aalice_devOP·2 days ago

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

6
Ttang_hao·2 days ago

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

152
Hhuang_ke·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.

65
Nnikic·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.

62
Sswoole_lee·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.

481
Kkernel_panic·12 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.

449
Rran_bo·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.

403
Oops_wang·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.

22
NnikicOP·2 days ago

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

343
Kkernel_panic·2 days agoedited

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

337
Hhuang_keOP·2 hours ago

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

248
Mmike_xu·yesterday

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

285
Zzhou_yi·2 days ago

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

280
Ttang_hao·2 days ago

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

3
Cchen_dev·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.

201
Oops_wangOP·2 days agoedited

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

99
Mmike_xu·2 hours 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.

158
Aalice_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.

148
Bbob_chen·2 days ago

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

148
Wwinter·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.

137
Bbob_chen·2 days ago

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

112
WwinterMod·2 days agoedited

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.

109
Rrase·28 minutes 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.

90
Aalice_dev·just now

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.

75
Oops_wang·12 minutes ago

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

150
Aalice_dev·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.

27
Ddev_zhou·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.

1
Ttang_hao·3 minutes ago

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

39
RraseOP·2 days ago

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

1
Zzhu_zong·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.

61
Sswoole_lee·5 hours 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.

50
Llinlin·12 minutes agoedited

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.

47
Wwinter·2 hours ago

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

41
Sswoole_lee·2 days ago

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

27
Ddev_zhou·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.

26
Aalice_dev·28 minutes agoedited

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

384
Lli_ming·2 days ago

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

78
Nnikic·2 days ago

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

95
Sslow_query·2 days ago

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

82
Cchen_devMod·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.

135
Ddev_zhou·2 days agoLevel 6

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

80
Bbob_chen·2 days ago

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

2
Kkite·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.

82
Wwinter·2 days ago

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

499
Lli_ming·2 days 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.

60
Zzhou_yi·5 hours ago

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

1
Rrase·3 minutes agoedited

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

1
Rrase·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.

83
Lli_ming·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.

25
Oops_wang·2 hours ago

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

14
Lli_ming·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.

9
Sslow_query·1 hour ago

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

3
Lli_mingMod·2 days ago

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

3
Sswoole_lee·just nowedited

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

1
Llinlin·1 hour ago

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

161
Bbob_chen·2 days ago

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

284
Bbob_chen·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.

153
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.

2
Rran_bo·just now

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.

1
Kkernel_panic·2 hours 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.

89
Rrase·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.

409
Zzhu_zong·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.

104
Mmike_xu·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.

12
Wwinter·yesterday

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.

23
Zzhou_yi·2 days ago

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

310
Aalice_dev·yesterdayedited

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

1

This is the post detail page /en/c/edge-computing-show/post/p8. 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 →