Add storage container docs (#294)

* Add storage container docs Signed-off-by: greenmoon55 <[email protected]> * Update storage container doc Signed-off-by: greenmoon55 <[email protected]> * Move doc to Model Storage section Signed-off-by: greenmoon55 <[email protected]> * Update docs/modelserving/storage/storagecontainers.md Signed-off-by: Dan Sun <[email protected]> * Add warning section Signed-off-by: Dan Sun <[email protected]> * Update storagecontainers.md Signed-off-by: Dan Sun <[email protected]> --------- Signed-off-by: greenmoon55 <[email protected]> Signed-off-by: Dan Sun <[email protected]> Co-authored-by: Dan Sun <[email protected]>
kserve · Oct 14, 2023 · 74693a7 · 74693a7
1 parent abdc5bd
commit 74693a7
Show file tree

Hide file tree

Showing 2 changed files with 65 additions and 0 deletions.
diff --git a/docs/modelserving/storage/storagecontainers.md b/docs/modelserving/storage/storagecontainers.md
@@ -0,0 +1,64 @@
+# Storage Containers
+
+KServe downloads models using a storage initializer (initContainer). For example, this is [the default storage initializer implementation](https://github.com/kserve/kserve/blob/79f2a48d0c9c72b034127170e38d6b29b927f03a/python/kserve/kserve/storage/storage.py). KServe introduced **ClusterStorageContainer** CRD in 0.11 which allows users to specify a custom container spec for a list of supported URI formats. 
+
+A _ClusterStorageContainer_ defines the container spec for one or more storage URI formats. Here is an example of a ClusterStorageContainer that corresponds to the default storage initializer. Note that this is incluced in the [helm chart](https://github.com/kserve/kserve/blob/79f2a48d0c9c72b034127170e38d6b29b927f03a/charts/kserve-resources/templates/clusterstoragecontainer.yaml). 
+
+```yaml
+apiVersion: "serving.kserve.io/v1alpha1"
+kind: ClusterStorageContainer
+metadata:
+ name: default
+spec:
+ container:
+ name: storage-initializer
+ image: kserve/storage-initializer:latest
+ resources:
+ requests:
+ memory: 100Mi
+ cpu: 100m
+ limits:
+ memory: 1Gi
+ cpu: "1"
+ supportedUriFormats:
+ - prefix: gs://
+ - prefix: s3://
+ - prefix: hdfs://
+ - prefix: webhdfs://
+ - regex: "https://(.+?).blob.core.windows.net/(.+)"
+ - regex: "https://(.+?).file.core.windows.net/(.+)"
+ - regex: "https?://(.+)/(.+)"
+```
+
+In a _ClusterStorageContainer_ spec, you can specify container resource requests and limits, and a list of supported URI formats that this image supports. KServe can match the URI either with `prefix` or `regex` .
+
+!!! warning
+
+ If a storage URI is supported by two or more _ClusterStorageContainer_ CRs, there is no guarantee which one will be used. **Please make sure that the URI format is only supported by one ClusterStorageContainer CR**.
+
+
+If you would like to use a custom protocol `model-registry://`, for example, you can create a custom image and add a new ClusterStorageContainer CR like this:
+
+```yaml
+apiVersion: "serving.kserve.io/v1alpha1"
+kind: ClusterStorageContainer
+metadata:
+ name: abc
+spec:
+ container:
+ name: storage-initializer
+ image: abc/custom-storage-initializer:latest
+ resources:
+ requests:
+ memory: 100Mi
+ cpu: 100m
+ limits:
+ memory: 1Gi
+ cpu: "1"
+ supportedUriFormats:
+ - prefix: model-registry://
+```
+
+## Spec Attributes
+
+Spec attributes are in [API Reference](/website/reference/api/#serving.kserve.io/v1alpha1.ClusterStorageContainer) doc.
diff --git a/mkdocs.yml b/mkdocs.yml
@@ -52,6 +52,7 @@ nav:
  - Concept: modelserving/inference_graph/README.md
  - Image classification inference graph: modelserving/inference_graph/image_pipeline/README.md
  - Model Storage:
+ - Storage Containers: modelserving/storage/storagecontainers.md
  - Azure: modelserving/storage/azure/azure.md
  - PVC: modelserving/storage/pvc/pvc.md
  - S3: modelserving/storage/s3/s3.md