Datalore Enterprise installation guide
This page is an overview of Datalore Enterprise installation options. You can skip it and advance directly to the installation topics. However, we recommend reading this page to get a better idea of the entire process.
What is Datalore Enterprise
Datalore Enterprise is a self-hosted version of Datalore, tailored for teams that want to have a collaborative data science platform set up on their own infrastructure.
Watch Datalore overview or schedule a personalized demo with our team to learn more about the product.
Licenses
You can choose:
Free license: for teams of 1-4 users, support provided on the public forum
Paid license: for teams of 5 or more users, support provided via email: datalore-enterprise@jetbrains.com
Installation process overview
The installation workflow comprises the following stages:
Infrastructure preparation and installation method selection
Before the installation of Datalore Enterprise itself, prepare the platform where you want to deploy it. Once the platform is prepared, select your installation method:
Refer to the table below to consider your infrastructure options and compare the two installation methods.
Docker deployment | Kubernetes deployment | ||
|---|---|---|---|
Platform | The following options are suitable for both methods:
| ||
General information | Docker Compose method This method is recommended unless you specifically want to use a Kubernetes cluster. If you have not decided which platform or service to choose yet, we recommend that you try installing Datalore Enterprise on an AWS EC2 instance. Refer to this guide to learn how to create your own infrastructure in AWS. | Helm charts method Besides creating and running a cluster directly on a virtual machine (eg. EC2 instance), you can use dedicated Kubernetes services: | |
Key differences |
When using several machines (Docker Swarm), you cannot control which one will be launched next. This limits your options when using GPU on your machines. You either have GPUs on all of them or restart computations until a GPU is launched. Kubernetes deployment allows you to control that. |
| |
Hardware requirements |
Because users have different needs and projects, we can only suggest a simplified formula for you to get a better idea of how calculating this value can be done: 4 * (N + 1) where N is the number of projected users, each working on one document.For example, for a team of 10 people working as described above, you will need at least 44 GB of RAM. | ||
System requirements |
|
This installation was tested with Kubernetes v1.24 and Helm v3.11.1, but other versions may work too. | |
Basic installation
The next stage is the basic installation procedure that allows you to get Datalore Enterprise up and running on the infrastructure of your choice. Follow one of the guides below to use the method you selected.
As a result, you will have Datalore Enterprise installed and ready for use, as well as your first signed-up Datalore user, who will have the administrator rights.
Configuration procedures
Whichever basic installation method you choose, it is followed by a set of configuration procedures.
The table below lists all configuration procedures. Unless noted otherwise, a procedure is applicable for both installation methods.
Procedure | Description |
|---|---|
Required | |
Configure Docker Swarm agents (for Docker deployment only) | Used to connect multiple machines to your Datalore installation |
Used to customize how your agents work to manage your computational resources | |
Used to enable GPU machines | |
Used to customize resource usage among your Datalore users | |
Optional | |
| Customize or update environment | Used to create multiple base environments out of custom Docker images |
Used to integrate an authentication service | |
Used to connect and configure agents outside your Datalore infrastructure | |
Used to enable a service generating and distributing gift codes | |
Used to activate email notifications | |
Used to set up auditing of your Datalore users | |
Upgrade your version
Each installation method has its own upgrade procedure:
We duly notify our customers of new version releases so they can promptly upgrade. Generally, upgrading a Docker-based installation is quicker and easier than the Kubernetes-based one.