Oracle Cloud Infrastructure
With Docker/Podman
Deploying Deepgram on Oracle Cloud Infrastructure (OCI) requires some preparation. In this section, you will learn how to provision a Virtual Machine where you will deploy Deepgram products. You will need to perform some of these steps in the Oracle Cloud Console and some in your local terminal.
Prerequisites
Make sure you have completed the requirements in the Self-Hosted Introduction.
Accessing Your Cloud Environment
Create an OCI Compute Instance
To begin your self-hosted installation with OCI as your cloud service provider, you need to create an OCI instance.
-
In the OCI Console, navigate to the Compute section. Select the compartment you want to deploy with. Click Instances, and then Create instance.
- Oracle recommends creating resources in compartments other than the
root
compartment on your IAM tenancy. If needed, see Oracle's documentation for details on how to create a new compartment.
- Oracle recommends creating resources in compartments other than the
-
For the Name wizard step, type
Deepgram Self-Hosted
, or another appropriate description of this deployment. -
For the Create in compartment wizard step, select which compartment you want to deploy this instance to.
-
For the Placement wizard step, the default options are generally appropriate. You may edit the options, including advanced options, according to your needs.
-
For the Security wizard step, edit the options according to your needs.
-
For the Image and shape wizard step, click Edit to choose a basic configuration to serve as a template for your instance:
-
Shape
-
Click Change shape to modify the underlying hardware for your instance.
-
Select the
Virtual machine
instance type. -
Select the
Speciality and previous generation
shape series to view shapes with GPUs available. -
Search and select
VM.GPU.A10.1
, or another GPU instance type that meets our minimum required hardware specifications.- If none of the shapes shown have an attached GPU, you may need to upgrade your access. The
Browse all shapes
window has a banner on the top offering you the option to upgrade. Follow the instructions to enroll in a plan or request a sales call to gain access to GPU compute.
- If none of the shapes shown have an attached GPU, you may need to upgrade your access. The
-
-
Image
- Pick an image from our list of recommended Linux distributions.
- Click Change Image, then search for and select your desired image.
Some images on OCI, such as Oracle Linux 8, have NVIDIA drivers built in. In the selected imaged drop-down, choose an
Image Build
that containsGPU
in the name.If you choose an image with NVIDIA drivers built in, you can skip the
Install NVIDIA Drivers
step in Drivers and Containerization Platforms.
-
-
For the Primary VNIC information wizard step, edit the options according to your needs.
- Under Primary VNIC IP addresses ensure that
Automatically assign public IPv4 address
is enabled. This will allow you to remotely login to your server.
If you want to be able to receive requests from the public internet, make sure to configure your Oracle subnet and internet gateway appropriately.
See Oracle's documentation on creating a VCN, creating a subnet , and creating an internet gateway for more details.
- Under Primary VNIC IP addresses ensure that
-
For the Add SSH keys wizard step, decide whether to generate a new key pair or use an existing SSH key. If you have Oracle generate a new key pair for you, download the
.key
file and take note of the location. -
For the Boot volume wizard step, change the boot volume size if necessary to meet our minimum required hardware specifications . Edit the other available options according to your needs.
If you increase the Boot volume size on instance creation or by editing a running instance, make sure to expand the root partition of the instance to use this additional allocated storage. See the Oracle documentation on
oci-growfs
for guidance on how to do this. -
For the Block volume wizard step, attach block volumes according to your needs.
-
Verify that the wizard contains all the right details for your desired instance. Click
Create
to create your compute instance. -
If you would like to save these options for future usage, click
Save stack
instead and follow the steps presented. Make sure to checkRun apply
during the review step to create an instance. -
The job details for the instance creation will be displayed. The Terraform plan and logs will show, and once completed, the instance creation job will be marked as
Succeeded
.
Log in to the OCI Compute Instance
To complete the rest of the installation, including configuring your environment and transferring files between your local computer and your OCI instance, you must connect to the instance that you launched.
-
You need to find the instance's Public IPv4 DNS entry. In the OCI Console, navigate back to the
Compute -> Instances
page. -
Find the newly provisioned compute instance. Take note of the Public IP assigned to this instance. This will be the
IP_ADDR
value that is referenced later. -
Open the terminal application for your computer.
-
Connect to your OCI instance:
ssh -i /path/to/private-key.key USERNAME@IP_ADDR
Be sure to replace the
IP_ADDR
andUSERNAME
placeholder values. TheIP_ADDR
is the public IP for your instance. If you used an Oracle Linux or CentOS platform image, theUSERNAME
isopc
. If you used an Ubuntu platform image to create the instance, the username isubuntu
.Also check that the path to your
private-key.key
file is correct.If you are on a Windows machine, PuTTY is an alternative to
ssh
. -
If you receive a message that indicates that the authenticity of the host can’t be established, type
yes
, then press theEnter
key on your keyboard. This message should only appear the first time you are connected to the instance.
Updated 6 months ago
Now that we have provisioned a deployment environment, we need to start configuring it for inference.