Skip to main content
Version: v0.9

Accelerate Different Storage with Fluid

We need to add additional configuration to Alluxio for accessing the storage normally, when using different underlying storage services as the underlying storage system of Alluxio.

This document shows how to declaratively complete the special configuration required by Alluxio in Fluid to access different storage services such as S3, HDFS, Ceph S3, PV, and MinIo. Please visit Amazon AWS S3 - Alluxio v2.8.1 (stable) Documentation for more information.

Prerequisites​

  • Before everything we are going to do, please refer to Installation Guide to install Fluid on your Kubernetes Cluster, and make sure all the components used by Fluid are ready like this:

    $ kubectl get pod -n fluid-system
    NAME                                  READY   STATUS    RESTARTS   AGE
    alluxioruntime-controller-5b64fdbbb-84pc6   1/1     Running   0          8h
    csi-nodeplugin-fluid-fwgjh                  2/2     Running   0          8h
    csi-nodeplugin-fluid-ll8bq                  2/2     Running   0          8h
    dataset-controller-5b7848dbbb-n44dj         1/1     Running   0          8h

  • The Storage which can be visited.

Configuration​

Create Dataset Resource Object

$ cat << EOF > dataset.yaml
apiVersion: data.fluid.io/v1alpha1
kind: Dataset
metadata:
  name: my-hdfs
spec:
  mounts:
    - mountPoint: hdfs://<namenode>:<port>/path1
      name: hdfs-file1
    - mountPoint: hdfs://<namenode>:<port>/path2
      name: hdfs-file2
EOF
$ kubectl create -f dataset.yaml

Fluid mounts the mountPoint attribute defined in the CRD object to Alluxio, so this attribute can be any legal UFS address that can be recognized by Alluxio.

In addition, multiple mountPoint can be set for each Dataset, so that you can mount all of the mountPoint to the specified directory. At the same time, you can also set subPath when mounting PVC to specify a mountPoint or its subdirectory set in the mount dataset. For example, in the above example, when mounting PVC to your pod, you can set subPath: hdfs-file1, so that only the hdfs://<namenode>:<port>/path1 directory will be mounted.

You can modify the spec.mounts field as required. It is generally set to the access path of the underlying storage, for example:

  • HDFS：- mountPoint: hdfs://<namenode>:<port>ï¼›

  • AWS S3：

    apiVersion: data.fluid.io/v1alpha1
    kind: Dataset
    metadata:
      name: my-s3
    spec:
      mounts:
        - mountPoint: s3://<bucket-name>/<path-to-data>/
          name: s3
          options:
            alluxio.underfs.s3.region: <s3-bucket-region>
            alluxio.underfs.s3.endpoint: <s3-endpoint>
            encryptOptions:
            - name: aws.accessKeyId
              valueFrom:
                secretKeyRef:
                  name: mysecret
                  key: aws.accessKeyId
            - name: aws.secretKey
              valueFrom:
                secretKeyRef:
                  name: mysecret
                  key: aws.secretKey

  • PVC：- mountPoint: pvc://nfs-imagenetï¼›

  • local path：- mountPoint: local:///mnt/nfs-imagenetï¼›

  • GCS：- mountPoint: gs://<bucket-name>/<path-to-data>

You need to specify the storage location in spec.mounts.mountPoint; In spec.mounts.options, specify the region, endpoint, and key required to access the storage（Refer to List of Configuration Properties - Alluxio v2.8.1 (stable) Documentation for more options）

Create AlluxioRuntime Resource Object

$ cat << EOF > runtime.yaml
apiVersion: data.fluid.io/v1alpha1
kind: AlluxioRuntime
metadata:
  name: my-hdfs
spec:
  ...
EOF

No additional configuration required in AlluxioRuntime for different underlying storage(Except HDFS，Please refer HDFS).

$ kubectl create -f runtime.yaml

So far, Alluxio can normally access different types of underlying storage according to the user specified configuration file.

Examples​