Your Browser is Out of Date

Nytro.ai uses technology that works best in other browsers.
For a full experience use one of the browsers below
Design Guide—Implementing a Digital Assistant with Red Hat OpenShift AI on Dell APEX Cloud Platform
Introduction Solution Concepts Solution Architecture and Requirements Solution Deployment
Introduction
Dell APEX Cloud Platform for Red Hat OpenShift
Object storage
Digital assistant deployment and configuration
Model serving environment
Data science project in Red Hat OpenShift AI
Project Workbench
LLM model preparation
Model deployment
Redis deployment
Document embeddings
Gradio deployment
Deployment summary
Solution execution Summary References Open-Source License

Redis deployment

Redis deployment

  • This section describes the Redis vector store deployment. Redis is an open source, in-memory data store that can be used as a Vector Store for your embeddings in a RAG scenario.

    Redis Deployment and Configuration (GitHub)

    1. From the OperatorHub, install the Redis Enterprise Operator. You can install the operator with the default value in the namespace that you want to create your Redis cluster (the operator is namespace based).

    2. Create new Project in OpenShift cluster:

    oc new-project redisdb

    3. Create Security Context Constraints using the following yaml manifests:

    apiVersion: security.openshift.io/v1
kind: SecurityContextConstraints
metadata:
   name: redis-enterprise-scc-v2
   annotations:
     kubernetes.io/description: redis-enterprise-scc is the minimal SCC needed to run Redis Enterprise nodes on Kubernetes.
      It provides the same features as restricted-v2 SCC, but allows pods to enable the SYS_RESOURCE capability,
      which is required by Redis Enterprise nodes to manage file descriptor limits and OOM scores for database shards.
      Additionally, it requires pods to run as UID/GID 1001, which are the UID/GID used within the Redis Enterprise node containers.
allowedCapabilities:
- SYS_RESOURCE
allowHostDirVolumePlugin: false
allowHostIPC : false
allowHostNetwork: false
allowHostPID: false
allowHostPorts: false
allowPrivilegeEscalation: false
allowPrivilegedContainer: false
readOnlyRootFilesystem: false
runAsUser:
   type: MustRunAs
   uid: 1001
fsGroup:
   type: MustRunAs
   ranges:
   - min: 1001
     max: 1001
seLinuxContext:
   type: MustRunAs
seccompProfiles:
- runtime/default
supplementalGroups:
   type: RunAsAny

    4. Provide the operator permissions for Redis Enterprise Operator and Cluster pods.

    oc adm policy add-scc-to-user redis-enterprise-scc system:serviceaccount:redisdb:redis-enterprise-operator
 oc adm policy add-scc-to-user redis-enterprise-scc system:serviceaccount:redisdb:rec

    5. Redis Enterprise Operator deployment is now available using OpenShift dashboard. For a visual walkthrough of Redis Enterprise Operator installation in Openshift cluster, see the Developer Redis page.

    6. Once the operator is deployed, create a Redis cluster. You can use the following YAML definition as an example (adapt to your needs regarding size, persistency, storageClass, and more).

    apiVersion: app.redislabs.com/v1
kind: RedisEnterpriseCluster
metadata:
  name: rec
spec:
  serviceAccountName: rec
  redisEnterpriseNodeResources:
    limits:
      cpu: '4'
      ephemeral-storage: 10Gi
      memory: 4Gi
    requests:
      cpu: '4'
      ephemeral-storage: 1Gi
      memory: 4Gi
  bootstrapperImageSpec:
    repository: registry.connect.redhat.com/redislabs/redis-enterprise-operator
  clusterCredentialSecretName: rec
  nodes: 3
  persistentSpec:
    enabled: true
    storageClassName: gp3-csi
    volumeSize: 20Gi
  createServiceAccount: true
  username: your_admin_username
  clusterCredentialSecretRole: ''
  podStartingPolicy:
    enabled: false
    startingThresholdSeconds: 540
  redisEnterpriseServicesRiggerImageSpec:
    repository: registry.connect.redhat.com/redislabs/services-manager
  redisEnterpriseImageSpec:
    imagePullPolicy: IfNotPresent
    repository: registry.connect.redhat.com/redislabs/redis-enterprise
  uiServiceType: ClusterIP
  clusterCredentialSecretType: kubernetes
  servicesRiggerSpec:
    databaseServiceType: 'cluster_ip,headless'
    serviceNaming: bdb_name
  services:
    apiService:
      type: ClusterIP


    7. Once the Redis cluster is ready, you can deploy a database to host the vector store. It is important to enable the search module and set enough memory to hold the initial index capacity. See the following steps for the example.

    8. Create the secret for the Redis database using following YAML definition as an example.

    apiVersion: v1
kind: Secret
metadata:
  name: redb-my-doc
  namespace: redisdb
type: Opaque
stringData:
  username: default
  password: 1a2b3c4d

     9. Create the Redis database using following YAML definition as an example.

                apiVersion: app.redislabs.com/v1alpha1
kind: RedisEnterpriseDatabase
metadata:
  name: my-doc
   namespace: redisdb
spec:
   databaseSecretName: redb-my-doc
   memorySize: 4GB
   modulesList:
    - name: search
       version: 2.8.4
   persistence: snapshotEvery12Hour
   replication: true
   tlsMode: disabled
  type: redis

    10. Once the database is deployed you will have:

    • A Secret named redb-my-doc (or the name you put in the YAML). It holds the password to the default user account for this database.
    • A Service named my-doc-headless (or the name you put in the YAML). From this Service you will get: The full URL to the service from within the cluster (such as my-doc-headless.my-namespace.svc.cluster.local) and the port that Redis is listening to (like 14155)

    11. Your Redis URL is in the different notebooks or applications within this repository, the full URI you can construct will be similar to: redis://default:password@server:port.

    With our example, it would be:

    redis://default:1a2b3c4d@my-doc-headless.my-namespace.svc.cluster.local:14155.