AI Pods

Ola Krutrim offers a state-of-the-art container service designed to meet the diverse computational needs of developers, researchers, and enterprises. Our platform provides a range of high-performance machine configurations, enabling users to select the optimal setup for their specific projects, whether for training complex machine learning models, running intensive simulations, or handling other GPU-accelerated tasks. AI Pod is more suitable for deploying a service, e.g., an inference endpoint. It is also suitable for development due to its Jupyter Notebook feature.

A Pod includes several components: a container volume that houses the operating system and temporary storage, a disk volume designated for permanent storage, an Ubuntu Linux container, assigned vCPU and system RAM, GPUs for specialized tasks, a pre-configured template to simplify software access, an SSH connection, and a proxy connection to enable web access.

Each Pod encompasses a variety of components:

  • A container disk that houses the operating system and temporary storage. This storage is volatile and will be lost if the Pod is stopped.

  • A volume disk for permanent storage, associated with the pod as long as it is not terminated. This storage is persistent and will be available even if the Pod is stopped. But if the pod is terminated, the volume disk will be terminated, and the content in the disk will be deleted.

  • An Ubuntu Linux container capable of running almost any software that can be executed on Ubuntu. We will add other OS flavors in the future. Please reach out to [email protected] if you want to request a new OS flavor.

  • Assigned vCPU and system RAM dedicated to the container and any processes it runs.

  • GPUs, tailored for specific workloads like CUDA or AI/ML tasks.

  • A pre-configured template that automates the installation of software and settings upon Pod creation, offering straightforward, one-click access to various packages.

Setting up a Pod

You need to select the right configuration of Pods available based on your requirements. You should focus on three items specifically, depending on the use case:

  • GPU

  • VRAM

  • Disk Size

We also support spliced GPUs. You can use it for use cases that don’t require a full GPU.

You can use the help of the tools below to understand Pod requirements:

Once you have a clear understanding of the pod configurations required, follow the steps below:

  1. Choose the configuration of the Pod you want to use.

  2. Please select the template you want installed in the pod specific to your use case. If you want any more templates installed from our side, please reach out to us at [email protected]

  3. Please add your SSH key. You can refer to this document on how to generate an SSH key. Please select the checkbox for SSH terminal access if you want to access the Pods through the terminal.

  4. Please select the checkbox for Jupyter Notebook if you want to run your code on Jupyter Notebook

  5. Please add the respective volume mount path to which you want to connect the volumes.

  6. Please select the Container Disk (Temporary) and Volume Disk (Persistent). Please note that you will be billed for the storage you selected and not for how much storage you used.

  7. Once all details are filled in, you can click on Deploy to deploy the pod.

  8. Once the pod is deployed, you can see the “Running” status on your pod.

Currently, we don’t support editing a Pod once it’s deployed. Once a pod is deployed with the container disk and volume disk selected, you can’t add more.

Connecting to a Pod

Follow the steps below to connect to a pod:

  1. Click on the three dots to the right of your pod

  2. Click on Connect

  3. You can use two ways to connect to the pod:

    • By clicking on the “Open” button near the Jupyter Notebook. This will open the Jupyter Notebook in a new tab, and you can run your code in there

    • Copy and paste the command mentioned in the SSH terminal in your terminal and run it in your system terminal. This will connect your terminal to the pod

Stopping and terminating a Pod

You can click on the Stop button to stop your Pod.

  • Please note that you won’t be charged for the Pod or container disk once your pod is stopped, but you will be charged for the Volume disk till the pod is terminated.

  • You will also lose all the data in the container disk once the pod is stopped.

  • Your Volume disk will remain intact, as well as the data in the Volume disk. You can access the data once you restart the pod.

You can click on the Terminate button to terminate your pod.

  • Once you terminate the pod, you won’t be charged for the pod, or the container disk, or the volume disk.

  • You will lose access to both the container disk and the volume disk. You will lose the data as well.

Last updated

Was this helpful?