Plixer ML engine

Deploying a machine learning engine means integrating machine learning models into various environments where they can be used to make predictions and decisions in real-time.

Deployment options vary and can include cloud-based platforms such as AWS and Azure, which offer scalability and ease of use, as well as local and on-premise solutions like vSphere, providing greater control and security.

Authentication token

Before anything else, a Plixer Scrutinizer auth token must be manually created. To do this, follow these steps:

  1. On the Plixer Scrutinizer web interface, navigate to Admin > Users & Groups > Authentication Tokens.

  2. Click the + button to add a new authentication token.

  3. Set the expiration date of the authentication token to One Year.

  4. Set the user to admin.

  5. Click Generate Token. Take note of the generated token. You’ll need this when running the setup.sh later in the setup process.

SSH credentials

During setup, the Plixer ML engine needs to be able to SSH into your Plixer Scrutinizer primary host. Take note of your primary IP address, as well as your plixer user’s SSH password.

If you require a private key for SSH, make sure that the public key is configured on your Plixer Scrutinizer at /home/plixer/.ssh/authorized_keys. Make the private key available on the machine that you are deploying the ML engine from. Take note of the path to the private key file, you’ll be asked for it during setup.sh.

Using an On-Premises Plixer Scrutinizer

Plixer Enterprise in AWS supports pairing with an On-Premises Plixer Scrutinizer, but a few network factors have to be considered.

VPC Site-to-Site connection

A VPC Site-to-Site connection can be used to enable communication between Plixer Enterprise in AWS and an On-Premises Plixer Scrutinizer. To configure a Site-to-Site connection between a VPC and a customer’s environment, follow the instructions here.

Direct Access via Public IP

An alternative to using a Site-to-Site connection is allowing external access to the On-Premises Plixer Scrutinizer using a public IP address.

In this case, the Plixer Scrutinizer instance in your environment will be exposed to external access via the internet. Ports 5432, 22, and 443 will be exposed. This public IP is what you will specify when running 01_aws_infrastructure.sh or setup.sh.

Take note of the internet gateway IP, because this will be manually added to the Plixer Scrutinizer pg_hba.conf file to allow access to Postgres. After adding the IP address to pg_hba.conf, run the command psql -c "SELECT pg_reload_conf()" on Plixer Scrutinizer to reload the configuration.

Setting environment variable on Plixer Scrutinizer

Important

This step applies only to version 19.4 of the Plixer ML engine. Future releases of the Plixer ML engine eliminates the need to set up the environment variable.

Before pairing the ML engine with Plixer Scrutinizer, the environment variable NO_ML_HEARTBEAT must be set. To do this, add the line export NO_ML_HEARTBEAT=1 to the local_env file.

$ cat ~/scrutinizer/env/local_env
### Add custom environment variable configuration directory that
### should not be overwritten on upgrade here.
### other values may be present here...
export OTHER_VARIABLE=1

export NO_ML_HEARTBEAT=1

To make sure that the variable is set:

$ echo $NO_ML_HEARTBEAT
1

Once the variable is set, follow the corresponding guide below to pair your ML engine with your Plixer Scrutinizer.