Deploy Graph Lakehouse with Helm

Follow the instructions below to deploy Graph Lakehouse using Helm.

1. Run the following command to add the Cambridge Semantics repository to Helm:

   helm repo add csi-helm https://storage.googleapis.com/csi-helm/

2. Run the following command to update the metadata for the Helm repository.

   helm repo update

3. Run the following command to find the available Graph Lakehouse Helm charts.

   helm search repo anzograph

   The command returns details about the two charts that are available, one for Graph Lakehouse and one for the Cambridge Semantics Apache Zeppelin image (see Use Third-Party Visualization Tools for information about the Zeppelin image).

   NAME CHART VERSION APP VERSION DESCRIPTION csi-helm/anzograph 2.0.20230427 3.1.5 CSI Anzograph deployment on K8S csi-helm/zeppelin 0.2.20191219 0.8.2 CSI Zeppelin deployment on K8s that enables...

4. Run the following command to fetch and view the readme for the Graph Lakehouse Helm chart:

   helm inspect readme csi-helm/anzograph | tee Readme.md

5. Run the following command to fetch and view the Graph Lakehouse Helm chart values (values.yaml):

   helm inspect values csi-helm/anzograph | tee values.yaml

6. By default the Helm chart is configured to deploy a single Graph Lakehouse node with 2 CPU and 8 GiB of RAM. If you want to customize the depolyment, such as to specify a larger instance or create a cluster, customize values.yaml before you deploy Graph Lakehouse. In addition, if you obtained a license key from Altair, add that key to values.yaml. The steps below provide guidance for customizing node or cluster sizes and adding a license key to the deployment. For more detailed information about all of the Graph Lakehouse Helm chart options, view the readme, Readme.md.

   1. Open values.yaml in a text editor. The file is in the $HELM_HOME directory that was defined when you initialized Helm, usually your home directory. You can run helm home to view the HELM_HOME location.

   2. The option that controls the number of instances for the cluster is in the Values for statefulset section of the file:

      # Values for statefulset replicas: 1

      To create a cluster, change the replicas value from 1 to the number of nodes that you want to deploy. To achieve the best performance, specify a multiple of 4, i.e., 4, 8, 12, etc. For guidance on sizing Graph Lakehouse servers and clusters, see Sizing Guidelines for In-Memory Storage.

   3. To increase the number of CPU or amount of memory on the instances that will be deployed, change the values for the cpu and memory settings under database.resources.requests. Depending on the values that you specify for requests, you might need to increase the values under limits.

      For example, the following values create a cluster with instances that have 16 CPU and 120 GiB of RAM each. The upper limit are instances with 32 CPU and 160 GiB.

      database: image: repository: "docker.io" name: "cambridgesemantics/anzograph-db" tag: "3.1.0" pullPolicy: "IfNotPresent" resources: requests: cpu: "16000m" memory: "120000Mi" limits: cpu: "32000m" memory: "160000Mi" tolerations: []

   4. When you have finished customizing the file, save and close values.yaml

7. Run the following command to deploy Graph Lakehouse:

   helm install -f ~/values.yaml <deployment_name> csi-helm/anzograph

   Where <deployment_name> is the unique name that you want to assign to this Graph Lakehouse deployment. For example:

   helm install -f ~/values.yaml anzograph-1 csi-helm/anzograph

   Helm deploys Graph Lakehouse and displays the initial status. For example:

   NAME: anzograph-1 LAST DEPLOYED: Fri May 12 22:32:12 2023 NAMESPACE: default STATUS: DEPLOYED

   RESOURCES: ==> v1/Pod(related) NAME READY STATUS RESTARTS AGE anzograph-anzograph-1-0 0/1 Pending 0 0s

   ==> v1/Secret

   NAME AGE anzograph-1-ui-secrets 1s anzograph-1-license 1s

   ==> v1/ConfigMap anzograph-1-configmap 1s

   ==> v1/Service anzograph-1-ui 1s anzograph-1-statefulset 1s

   ==> v1beta1/StatefulSet anzograph-anzograph-1 1s

8. Run the following command to refresh the status and monitor the deployment:

   helm status <deployment_name>

   For example:

   helm status anzograph-1

   When the status says "Running," the deployment is complete. In the status output under v1/Service, note the first service name (with -ui appended to the release name). In the example above, the service name is anzograph-1-ui.

9. Using the service name for your deployment, run the following command to view the cluster and endpoint information for Graph Lakehouse:

   kubectl get service <service_name>

   For example:

   kubectl get service anzograph-1-ui

   NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE anzograph-1-ui LoadBalancer 10.47.254.111 35.225.23.113 443:30281/TCP,80:30704/TCP 1h

For next steps, see Get Started for brief tutorials that are designed to introduce you to the Graph Lakehouse user interface and CLI and get you started with loading data and running SPARQL queries.