Automation for IBM Watson Deployments

Automation for IBM Watson Deployments (this repo) contains samples and documentation how to create managed OpenShift clusters on the IBM Cloud, how to deploy Watson NLP containers and how to access them from custom applications. With an IBM Cloud account a full production sample can be deployed in roughly one hour.

🔴 IMPORTANT: At this point this functionality is experimental only and subject to change.

Key Technologies:

IBM Watson NLP (Natural Language Processing) comes with a wide variety of text processing functions, such as emotion analysis and topic modeling. Watson NLP is built on top of the best AI open source software. Additionally it provides stable and supported interfaces, it handles a wide range of languages and its quality is enterprise proven. The cloud-native Watson NLP containers can be deployed on various Kubernetes-based platforms.
IBM Technology Zone Deployer is an opinionated open source framework on top of the standard technologies Terraform and ArgoCD. The toolkit allows automating deployments and operations of infrastructure like Red Hat OpenShift clusters and applications.

Happy Path

The steps below describe the "happy path" for developers and operation engineers to deploy a simple Watson NLP scenario on OpenShift as easily as possible. At the end of the happy path the following infrastructure, service, and application components will have been deployed:

Managed OpenShift cluster on the IBM Cloud on VPC (screenshot)
ArgoCD running in the OpenShift cluster (screenshot)
Watson NLP pod (screenshot)
- Runtime container which provides gRCP and REST interfaces via a service
- One model container for syntax predictions
Sample consumer application pod (screenshot)
- UBI container which is configured to run a command which invokes the Watson NLP runtime container via REST
Snippets to invoke Watson NLP via gRCP and REST from a local environment (screenshot)

The following flip book contains screenshots that describes the setup as well as the results.

Outline

Setup

The following parts describes the prerequisites, setup instructions and basic usage and validation.

Prerequisites

You need the following prerequisites:

Docker
IBM id and IBM Cloud (paid) account
iascable CLI (toolkit CLI)
Watson entitlement key (you can use a trial)
GitHub.com access token
grcpcurl
oc

1. Docker

Docker is needed for the instructions below. Alternatively Multipass can be used (see documentation).

2. IBM id and IBM Cloud (paid) account

Register on IBM Cloud and enter your payment information. Next create an IBM Cloud API key.

3. `iascable` CLI

iascable is the IBM Technology Zone Deployer CLI. Install iascable 2.25.4 or later.

curl -sL https://iascable.cloudnativetoolkit.dev/install.sh | sh
iascable --version

If you encounter the error zsh: killed on Apple M1, this is due to an issue with the iascable binary. As a temporary workaround, install iascable using an alternative approach.

If not already installed, install npm:

brew install npm

Install iascable via npm:

sudo -i rm /usr/local/bin/iascable
npm install -g @cloudnativetoolkit/iascable

4. Watson NLP trial entitlement key

An entitlement key is needed.

5. GitHub.com access token

Create an access token on GitHub.com.

6. `grcpurl`

Install grcpurl.

7. `oc`

Install oc.

Instructions

The following steps need to be performed:

Clone repo
Understand the BOM (Bill of Materials)
Define variables.yaml
Define credentials.properties
Run iascable build
Apply Terraform modules

1. Clone repo

git clone https://github.com/ibm/watson-automation
cd watson-automation/roks-new-nlp

2. Understand BOM (Bill of Materials)

The bill of materials (bom.yaml) contains the Technology Zone Deployer modules that will be deployed: OpenShift, ArgoCD, Watson NLP and an UBI container.

Note: You dont have to change this file.

spec:
  modules:
    - name: gitops-repo
      alias: gitops_repo
      version: v1.23.1
    - name: argocd-bootstrap
      alias: argocd-bootstrap
      version: v1.12.0
    - name: gitops-namespace
      alias: namespace
      version: v1.14.0
    - name: ibm-resource-group
      alias: resource_group
      version: v3.3.5
    - name: ibm-object-storage
      alias: cos
      version: v4.1.0
    - name: ibm-ocp-vpc
      alias: cluster
      version: v1.16.3
    - name: ibm-vpc
      alias: ibm-vpc
      version: v1.17.0
    - name: ibm-vpc-gateways
      alias: ibm-vpc-gateways
      version: v1.10.0
    - name: ibm-vpc-subnets
      alias: ibm-vpc-subnets
      version: v1.14.0
    - name: olm
      alias: olm
      version: v1.3.5
    - name: sealed-secret-cert
      alias: sealed-secret-cert
      version: v1.0.1
    - name: terraform-gitops-ubi
      alias: terraform_gitops_ubi
      version: v0.0.26
    - name: terraform-gitops-watson-nlp
      alias: terraform_gitops_watson_nlp
      version: v1.0.0
    - name: util-clis
      alias: util-clis
      version: v1.18.1

The Bill of Material can be used by different consumers. Consumer specific variables need to be defined in variables.yaml and credentials.properties.

3. Define `variables.yaml`

Create output/cluster-with-watson-nlp/variables.yaml based on output/cluster-with-watson-nlp/variables-template.yaml.

cp output/cluster-with-watson-nlp/variables-template.yaml output/cluster-with-watson-nlp/variables.yaml
vi output/cluster-with-watson-nlp/variables.yaml

Define the variables in variables.yaml. If not done, you will get prompted later. Replace all "xxx" occurances. You must also change "terraform_gitops_watson_nlp_accept_license" from false to true.

variables:
  # overall
  - name: region
    description: The IBM Cloud region where the instance should be provisioned
    value: xxx
  - name: resource_group_name
    description: The name of the IBM Cloud resource group where the resources should be provisioned
    value: xxx
  - name: common_tags
    description: The list of tags that should be applied to all resources (does not work)
    value: []

  # vpc
  - name: ibm-vpc-subnets__count
    description: The number of subnets that should be created in the VPC
    value: 1

  # ocp
  - name: worker_count
    description: The number of workers that should be provisioned per subnet
    value: 2
  - name: cluster_flavor
    description: The flavor of the worker nodes that will be provisioned
    value: bx2.4x16

  # namespace
  - name: namespace_name
    description: The name of the namespace that will be created within the cluster
    value: watson-demo

  # gitops
  - name: gitops_repo_repo
    description: The name of the gitops repository that will be created
    value: xxx
  - name: gitops_repo_host
    value: github.com
  - name: gitops_repo_org
    value: xxx
  - name: gitops_repo_username
    value: xxx

  # nlp
  - name: terraform_gitops_watson_nlp_runtime_image
    value: watson-nlp-runtime:1.0.18
  - name: terraform_gitops_watson_nlp_runtime_registry
    value: watson
  - name: terraform_gitops_watson_nlp_accept_license
    value: false
  - name: terraform_gitops_watson_nlp_imagePullSecrets
    value:
      - ibm-entitlement-key
  - name: terraform_gitops_watson_nlp_models
    value:
      - registry: watson
        image: watson-nlp_syntax_izumo_lang_en_stock:1.0.6
  - name: terraform_gitops_watson_nlp_registries
    value:
      - name: watson
        url: cp.icr.io/cp/ai
  - name: terraform_gitops_watson_nlp_registryUserNames
    value:
      - registry: watson
        userName: cp

  # ubi
  - name: terraform_gitops_ubi_command
    value: echo helloworld

4. Define credentials.properties

Create output/credentials.properties based on output/credentials-template.properties.

cp output/credentials-template.properties output/credentials.properties
vi output/credentials.properties

You need to define three credentials:

GitHub.com access token
IBM Cloud API key
Watson entitlement key

export TF_VAR_gitops_repo_token=xxx
export TF_VAR_ibmcloud_api_key=xxx
export TF_VAR_terraform_gitops_watson_nlp_registry_credentials=xxx

5. Run `iascable build`

Use iascable to create the Terraform code.

cd watson-deployments/roks-new-nlp
iascable build -i bom.yaml
cd output

6. Apply Terraform modules

An image is provided which comes with all tools necessary to run Terraform. To start the container, invoke this command in a terminal:

./launch.sh

In the running container invoke these commands:

cd cluster-with-watson-nlp
./apply.sh

Note: Be patient. Running apply.sh takes roughly 45 minutes.

🔴 IMPORTANT: Timing issues can occur and you get errors. In this case wait a couple of minutes and run the same script again. In some cases you need three attempts.

Usage

Once the Watson NLP pod and the UBI pods are running, gRPC and REST can be used to run predictions using the simple syntax model.

Log in to OpenShift and forward port(s). Execute the following commands in a first terminal window. Set the port to 8080 for REST or 8085 for gRPC.

ibmcloud login --sso
ibmcloud target -g RESOURCE_GROUP # e.g. ibmcloud target -g niklas
ibmcloud target -r REGION # e.g. ibmcloud target -r eu-gb
ibmcloud ks cluster ls
ibmcloud oc cluster config -c CLUSTER --admin # e.g. ibmcloud oc cluster config -c niklas-cluster --admin

or

oc login --token=sha256~xxx --server=https://xxx

To get the OpenShift login token, open the OpenShift console, click on your user name in the upper right corner and choose "copy login command".

YOUR_NAMESPACE=watson-demo
WATSON_SERVICE=$(oc get services -n $YOUR_NAMESPACE | grep "watson" |  awk '{print $1;}')
oc port-forward svc/$WATSON_SERVICE -n $YOUR_NAMESPACE 8085

gRPC

Execute the following command in a second terminal window.

git clone https://github.com/IBM/ibm-watson-embed-clients
cd ibm-watson-embed-clients/watson_nlp/protos
grpcurl -plaintext -proto common-service.proto -d '{
"raw_document": {
"text": "It is so easy to embed Watson NLP in applications. Very cool"
},
"parsers": ["token"]
}' -H 'mm-model-id: syntax_izumo_lang_en_stock' localhost:8085 watson.runtime.nlp.v1.NlpService.SyntaxPredict

Expected output:

{
  "text": "It is so easy to embed Watson NLP in applications. Very cool",
  "producerId": {
    "name": "Izumo Text Processing",
    "version": "0.0.1"
  },
  "tokens": [
    {
      "span": {
        "end": 2,
        "text": "It"
      }
    },
...
  ],
  "sentences": [
    {
      "span": {
        "end": 50,
        "text": "It is so easy to embed Watson NLP in applications."
      }
    },
    {
      "span": {
        "begin": 51,
        "end": 60,
        "text": "Very cool"
      }
    }
  ],
  "paragraphs": [
    {
      "span": {
        "end": 60,
        "text": "It is so easy to embed Watson NLP in applications. Very cool"
      }
    }
  ]
}

REST

Execute the following command in another terminal window.

curl -X POST "http://localhost:8080/v1/watson.runtime.nlp.v1/NlpService/SyntaxPredict" \
  -H "accept: application/json" \
  -H "grpc-metadata-mm-model-id: syntax_izumo_lang_en_stock" \
  -H "content-type: application/json" \
  -d " { \"rawDocument\": { \"text\": \"It is so easy to embed Watson NLP in applications. Very cool.\" }}"

Expected output:

{
  "text": "It is so easy to embed Watson NLP in applications. Very cool.",
  "producerId": { "name": "Izumo Text Processing", "version": "0.0.1" },
  "tokens": [
    {
      "span": { "begin": 0, "end": 2, "text": "It" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 3, "end": 5, "text": "is" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 6, "end": 8, "text": "so" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 9, "end": 13, "text": "easy" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 14, "end": 16, "text": "to" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 17, "end": 22, "text": "embed" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 23, "end": 29, "text": "Watson" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 30, "end": 33, "text": "NLP" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 34, "end": 36, "text": "in" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 37, "end": 49, "text": "applications" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 49, "end": 50, "text": "." },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 51, "end": 55, "text": "Very" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 56, "end": 60, "text": "cool" },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    },
    {
      "span": { "begin": 60, "end": 61, "text": "." },
      "lemma": "",
      "partOfSpeech": "POS_UNSET",
      "dependency": null,
      "features": []
    }
  ],
  "sentences": [
    {
      "span": {
        "begin": 0,
        "end": 50,
        "text": "It is so easy to embed Watson NLP in applications."
      }
    },
    { "span": { "begin": 51, "end": 61, "text": "Very cool." } }
  ],
  "paragraphs": [
    {
      "span": {
        "begin": 0,
        "end": 61,
        "text": "It is so easy to embed Watson NLP in applications. Very cool."
      }
    }
  ]
}

Open Swagger.

Resources

IBM Watson NLP
IBM Technology Zone Deployer
- Documentation
- Modules
- CLI
- Watson NLP Module
- UBI Module
- Labs
Embeddeable AI
IBM Technology Zone assets

License

This project is licensed under the Apache 2 License - see the LICENSE file for details.

Name		Name	Last commit message	Last commit date
Latest commit History 56 Commits
documentation		documentation
helm-nlp		helm-nlp
roks-existing-nlp		roks-existing-nlp
roks-new-nlp		roks-new-nlp
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Automation for IBM Watson Deployments

Happy Path

Outline

Setup

Prerequisites

1. Docker

2. IBM id and IBM Cloud (paid) account

3. `iascable` CLI

4. Watson NLP trial entitlement key

5. GitHub.com access token

6. `grcpurl`

7. `oc`

Instructions

1. Clone repo

2. Understand BOM (Bill of Materials)

3. Define `variables.yaml`

4. Define credentials.properties

5. Run `iascable build`

6. Apply Terraform modules

Usage

gRPC

REST

Resources

License

About

Contributors 10

License

IBM/watson-automation

Folders and files

Latest commit

History

Repository files navigation

Automation for IBM Watson Deployments

Happy Path

Outline

Setup

Prerequisites

1. Docker

2. IBM id and IBM Cloud (paid) account

3. iascable CLI

4. Watson NLP trial entitlement key

5. GitHub.com access token

6. grcpurl

7. oc

Instructions

1. Clone repo

2. Understand BOM (Bill of Materials)

3. Define variables.yaml

4. Define credentials.properties

5. Run iascable build

6. Apply Terraform modules

Usage

gRPC

REST

Resources

License

About

Topics

Resources

License

Stars

Watchers

Forks

Contributors 10

3. `iascable` CLI

6. `grcpurl`

7. `oc`

3. Define `variables.yaml`

5. Run `iascable build`