ocicdn
Free OCI registry pull-through cache. Last 14 days.
source — hero: AE sampled counts; tables below: Analytics Engine (sampled). Cache hit rate is full-pipeline (CF edge + regional R2 + central R2).
by cache tier
| tier | requests | bytes* |
|---|---|---|
EDGE-ENAM | 7,570 | 2.03 GB |
R2-ENAM | 290 | 215.35 MB |
MISS-ENAM | 128 | 59.33 MB |
UPSTREAM-ENAM | 60 | 0 B |
NONE | 16 | 1.32 KB |
MISS-APAC | 10 | 3.60 GB |
R2-CENTRAL | 1 | 25.69 MB |
* Bytes by tier are AE-sampled and count only response bodies
(HEAD responses don't contribute). These should sum to roughly the
hero's "bytes served" modulo AE sampling; any residual gap is
UPSTREAM-* traffic CF counts but isn't cache-classified.
by region
| region | continent | requests |
|---|---|---|
ENAM | NA | 8,048 |
APAC | AS | 10 |
CENTRAL | AS | 1 |
by registry
| registry | requests | pulls |
|---|---|---|
docker | 1,482 | 24 |
gcr | 1,434 | 5 |
quay | 1,306 | 11 |
ghcr | 1,291 | 9 |
k8s | 1,291 | 6 |
mcr | 1,271 | 10 |
"pulls" counts manifest-HEAD-by-tag events (one per docker pull);
"requests" is every /v2/ hit including blobs, HEADs, manifest GETs, etc.
by continent
| continent | requests |
|---|---|
NA | 8,064 |
AS | 11 |
top pulled images
| registry | image | pulls |
|---|---|---|
docker | library/alpine:3.20 | 14 |
quay | prometheus/node-exporter:latest | 11 |
mcr | azurelinux/distroless/minimal:3.0 | 10 |
ghcr | anchore/syft:latest | 9 |
k8s | pause:3.9 | 6 |
gcr | distroless/static:latest | 5 |
docker | library/alpine:.INVALID_MANIFEST_NAME | 3 |
docker | library/busybox:1.37 | 2 |
docker | library/python:3.12-slim | 1 |
Ranked by manifest-HEAD-by-tag events — each corresponds to one docker pull image:tag initiated by a user. Digest-pinned pulls (@sha256:…) are excluded to avoid double-counting the per-pull flow.