Download and Install Kinetica

This tutorial will walk you through how to download and install Kinetica. It’s a quick process, and you should be done in a few minutes.

License Key

No Key, No account
Kinetica requires a license key to activate the product.  Please register for your key.

Important

If using a trial license of Kinetica, note that a cluster of at most three nodes can be created.

Kinetica System Requirements

The base system requirements for Kinetica can be found in the table below. Step 1 lists the supported OS version numbers.

CPU Two-socket based server with at least 8 cores Intel x86-64 or Power PC 8le
GPU (Optional) Kinetica supports NVIDIA GPUs. We recommend K20, K40, K80, P100, GTX Ti780, Tegra, or similar.
Operating System
  • RHEL/CentOS
  • Debian/Ubuntu
  • SUSE
Memory Kinetica is an in-memory database. You’ll need at least enough system RAM to hold your data plus server resources. We recommend at least 8GB RAM.

If you plan on using the Active Analytics Workbench (AAW) extensively, you’ll need between 16 and 32 GB RAM.

Disk We recommend at least 4x RAM – although actual amount depends on workload, swap, etc.

Step 1: Download and Launch KAgent

KAgent can install and provision Kinetica using three different paths.

  1. KAgent installs Kinetica to the same host; other already-provisioned hosts can be added to the cluster.
  2. KAgent installs Kinetica to an already-provisioned host (or set of hosts) from cloud providers like AWS.
  3. KAgent installs Kinetica to new hosts it automatically provisions from those same providers.

We recommend starting with new, clean servers when following one of the already-provisioned host paths. Feel free to contact us for any questions. When you’re done installing, come back and join us for a Walkthrough of Kinetica.

KAgent can be installed through package managers for the following OS types.

Platform x86 ppc64le
Debian 8.x
Ubuntu 14.x LTS, 16.x LTS 14.04 LTS, 16.x LTS
RHEL 7.x 7.2
CentOS 7.x 7.x
SUSE 12, 12 SP1, 12 SP2

On Debian/Ubuntu:

Execute the below commands on your Ubuntu machine from a shell prompt, according to your server platform.

  1. Download the key file:

    wget -O - http://repo.kinetica.com/debian/7.0/Ubuntu/xenial/binary-amd64/keyFile | sudo apt-key add -
  2. Add the KAgent repo to the sources list:

    sudo sh -c 'echo "deb http://repo.kinetica.com/debian/7.0/Ubuntu/xenial/binary-amd64 ./" >> /etc/apt/sources.list'
  3. Update the package list:

    sudo apt update
  4. Install KAgent:

    sudo apt install kagent

  1. Download the key file:

    wget -O - http://repo.kinetica.com/debian/7.0/Ubuntu/xenial/binary-ppc64el/keyFile | sudo apt-key add -
  2. Add the KAgent repo to the sources list:

    sudo sh -c 'echo "deb http://repo.kinetica.com/debian/7.0/Ubuntu/xenial/binary-ppc64el ./" >> /etc/apt/sources.list'
  3. Update the package list:

    sudo apt update
  4. Install KAgent:

    sudo apt install kagent

On RHEL/CentOS:

Execute the below commands on your RHEL/CentOS machine from a shell prompt, according to your server platform.

  1. Download the repo file:

    sudo wget -O /etc/yum.repos.d/kinetica-7.0.repo http://repo.kinetica.com/yum/7.0/CentOS/7/x86_64/kinetica-7.0.repo
  2. Update the package list:

    sudo yum update
  3. Install KAgent:

    sudo yum install kagent

  1. Download the repo file:

    sudo wget -O /etc/yum.repos.d/kinetica-7.0.repo http://repo.kinetica.com/yum/7.0/CentOS/7/ppc64le/kinetica-7.0.repo
  2. Update the package list:

    sudo yum update
  3. Install KAgent:

    sudo yum install kagent

On SUSE:

Execute the below commands on your SUSE machine from a shell prompt.

  1. Add the repo file:

    sudo zypper addrepo http://repo.kinetica.com/yum/7.0/SLES/12sp2/x86_64/kinetica-7.0.repo
  2. Update the package list:

    sudo zypper update
  3. Install KAgent:

    sudo zypper install kagent



This installs the package to the directory /opt/gpudb/kagent, exposes port 8081, and registers the kagent service.

Step 2: Install and Configure Kinetica

Now that KAgent has been installed, you should be able to access the KAgent user interface to configure and start the service.

Note that this tutorial skips SSL and External Authentication setup in the Security section, but both are strongly recommended in a production environment. Additional security configuration can be made at any time using KAgent. Consult the KAgent documentation for more information.

KAgent should be running on port 8081. Open a browser and enter the following URL, making sure to insert the correct hostname or IP.

  1. Navigate to KAgent (http://<kagent-host>:8081/)
  2. Click Add New or Existing Cluster to begin the installation process.

Configuration

  1. Enter kinetica for the Name.
  2. For the Variant, select an install type: CPU-only (intel) or GPU (cuda80 or cuda91).

    If unsure which architecture to use, input the uname -a command into a shell local to the machine to determine the CPU architecture.

  3. Provide your License Key.

    License Key

    No Key, No account
    Kinetica requires a license key to activate the product.  Please register for your key.
  4. Select to install AAW (the Active Analytics Workbench) and KAgent.
  5. Select Automatic for the K8 Setup. This will automatically install Kubernetes for use with AAW.
  6. Click Next.

Deployment

Select a deployment provider.

Use the On Premise instructions if deploying to already-provisioned cloud hosts.

 


On premise requires no additional setup. KAgent will automatically open required ports for Kinetica services and functionality. Click Next.

  1. Provide your AWS account’s Access Key and Secret key. If you do not have one, create one in your AWS IAM console.
  2. Provide your AWS account’s SSH Key Name.
  3. Select an AWS Region.
  4. Click Next.

  1. Provide the GCP Project name.
  2. Upload a GCP Service Account File (typically a JSON file).

    The service account selected must have several compute network and firewalls permissions. We recommend assigning the Compute Admin role to the service account.

  3. Select a GCP Zone.
  4. Provide an SSH User name and SSH Public Key. The user and key will be added to the new cluster.
  5. Click Next.

  1. Provide an Azure application’s Client ID, Secret, Subscription ID, and Tenant.

    For assistance in finding these values and prepping the application for Kinetica installation, refer to the Azure documentation. We recommend assigning the application to the Contributor role.

  2. Select an Azure Region.
  3. Provide an SSH User name and SSH Public Key. The user and key will be added to the new cluster.
  4. Click Next.

Security

  1. Provide and confirm an Admin Password.

    This password will be used to access the administration account for each component in Kinetica (GAdmin, KAgent, AAW, and Reveal)

  2. Leave the default settings for SSL Mode and Authentication. Additional security configuration can be made post-installation using KAgent. Consult the KAgent documentation for more information.
  3. Click Next.

Nodes

Node setup will be different depending on the deployment provider you chose in Deployment.

Use the On Premise instructions if deploying to already-provisioned cloud hosts.

 

  1. Click + Add New Node until the amount of desired nodes for your cluster is reached.
  2. Provide the Hostname, Internal IP address, and Public IP address for each node.
  3. Click Next.
  4. Select if KAgent should attempt to connect to your cluster via Internal or Public IP. KAgent must be able to connect to your cluster to install Kinetica and its components. This is typically done via the Public IP address but in some network environments, the Internal IP address may be required.

  1. Click + Add New Node until the amount of desired nodes for your cluster is reached.
  2. Select an AWS Instance Type for each node.

    If you selected a GPU variant in Configuration, you should use a Accelerated Computing instance, like the p2, p3, or g3 series instance types. If you selected a CPU-only variant in Configuration, you should use a General Purpose instance, like the t2 series instance types.

  3. Select the Data Size (in GB) for each node. The size must be at least 32GB per node.
  4. Click Next.

  1. Click + Add New Node until the amount of desired nodes for your cluster is reached.
  2. Select a GCP Instance Type for each node.
  3. If you selected a GPU variant in Configuration, select nvidia-tesla-p4 for the GPU Card on each node. If you selected a CPU-only variant, we recommend selecting no-gpu for the GPU Card on each node.
  4. Select the Data Size (in GB) for each node. The size must be at least 32GB per node.
  5. Click Next.

  1. Click + Add New Node until the amount of desired nodes for your cluster is reached.
  2. Select an Azure Instance Type for each node.

    If you selected a GPU variant in Configuration, you should use a GPU Optimized instance, like the NC, ND, or NV series instance types. If you selected a CPU-only variant in Configuration, you should use an General Purpose instance, like the Standard_DS5_v2 instance type.

  3. Select the Data Size (in GB) for each node. The size must be at least 32GB per node.
  4. Click Next.

Credentials

Credentials configuration will be different depending on the deployment provider you chose in Deployment.

Use the On Premise instructions if deploying to already-provisioned cloud hosts.

 

  1. Provide an SSH Username and SSH Password OR SSH Private Key. These credentials will be verified to provide KAgent access to the machine(s).
  2. Optionally, provide a Sudo Password.
  3. Click Verify.

  1. Upload the SSH Private Key (AWS PEM) that matches the SSH Public Key you provided in Deployment.
  2. Optionally, provide a Sudo Password.
  3. Click Next.

  1. Input an SSH Username. It must match the SSH User you provided in Deployment.
  2. Upload the SSH Private Key that matches the SSH Public Key you provided in Deployment.
  3. Optionally, provide a Sudo Password.
  4. Click Next.

  1. Input an SSH Username. It must match the SSH User you provided in Deployment.
  2. Upload the SSH Private Key that matches the SSH Public Key you provided in Deployment.
  3. Optionally, provide a Sudo Password.
  4. Click Next.

Installation

Review the summary information, then click Install. Depending on the deployment and OS type, KAgent will ask permission to disable SELinux on the nodes. Kubernetes cannot be automatically installed if SELinux is not disabled. Click I Agree to continue with the installation.

It’s important that the browser window open with KAgent installing Kinetica remain open. If the browser closes at any point, the installation process will be cancelled but can be restarted later by reopening a browser window to KAgent.

A console will appear, logging every action taken to install Kinetica, AAW, and KAgent. The installation may take a while as KAgent adds a repository, downloads the package, installs the package to the directory /opt/gpudb, creates a group named gpudb, and two users (gpudb & gpudb_proc) whose home directories are located in /home/gpudb. This will also register four services on the new Kinetica instances: gpudb, gpudb_host_manager, kagent_ui, and kml (for AAW).

Next: Walkthrough of Kinetica »

Kinetica Trial Feedback