Kubeflow as a Service FAQ

How much does Kubeflow as a Service cost?

Arrikto Kubeflow as a Service comes with a 7-day free trial. After your free trial expires, you can upgrade to a paid plan. Cost starts from $2.06/hour per running Kubeflow deployment and $0.20/hour per stopped Kubeflow deployment.

Do I need a credit card to sign up and use the service?

No credit is needed to get started with Kubeflow as a Service.

How long is the free trial?

Kubeflow as a Service is completely free for the first 7 days. After that, you can upgrade to a paid plan.

How many Kubeflow deployments can I create?

You can create one Kubeflow deployment during the free trial. Upgrade to a paid plan and enjoy unlimited deployments.

How long does a Kubeflow deployment live?

You can use your Kubeflow deployments 24/7. Remember to use the stop button when you are not using your deployments.

I loved the trial, can I keep using the service?

Sure! Within the Kubeflow as a Service dashboard, click the UPGRADE button to upgrade to a paid plan and keep using the service.

I am a beginner with Kubernetes, can I still use Arrikto’s Kubeflow as a Service?

Absolutely! Kubeflow as a Service makes it easy by abstracting all the traditional Kubernetes complexities allowing you to focus on your ML projects.

How difficult is it to create a Kubeflow deployment using Kubeflow as a Service?

It requires a single click of a button. Click the button and a cluster will get created and be ready for you to use.

How much time does it take to create a Kubeflow deployment via Kubeflow as a Service?

There’s a lot of provisioning that takes place, which includes both the underlying infrastructure, Kubernetes, and the installation and configuration of Kubeflow, as well as testing that everything works as it should. So it can take up to 30 minutes for your initial Kubeflow deployment. Don’t worry though, since we have great educational material for you until the deployment is ready.

I need to create a multi-tenant Kubeflow cluster. Is it possible?

Absolutely! Contact us and schedule a meeting with a Kubeflow specialist.

 I want to run Kubeflow in production. What options do I have?

Arrikto can help you run Kubeflow in production in your preferred environment. Contact us and schedule a meeting with a Kubeflow specialist.

I think I did everything right but I’m stuck, can someone help me?

If you need assistance, click on Support from the left sidebar of Kubeflow as a Service, and we will try to get back to you as soon as we can.

Where can I learn more about Kubeflow? And are there model templates I can use?

Arrikto has an extensive collection of educational content, which can be accessed at academy.arrikto.com. As for models and templates, stay tuned!

My deployment is ready. How do I login?

You can find your deployment link and credentials under the LOGIN INFO column inside your Kubeflow as a Service dashboard.

My deployment is ready. Where do I start?

Go to academy.arrikto.com to discover a great collection of tutorials and courses that will help you get started with Kubeflow.

I see a deleted deployment. Can I restart it?

Unfortunately, no. But if you need more time with Kubeflow as a Service, you can extend the free trial from within the Kubeflow as a Service dashboard.

I’m using my email in the Kubeflow dashboard login page, but I can’t login, what am I doing wrong?

You can find your login information under the LOGIN INFO column inside your Kubeflow as a Service dashboard.

My deployment seems very slow, what is happening?

If you have some ML use cases that require more cloud resources, contact us or log a ticket and we will get back to you as soon as possible.

I need to use a GPU, but I can’t find such an option.

The Kubeflow as a Service doesn’t support GPUs yet. But Kubeflow does! So, if you have GPU needs, contact us and schedule a meeting with a Kubeflow specialist.

My deployment went into an error state while it was being created. What should I do? Does this mean I lost one out of three deployments?

Unfortunately, this has to do with how long the underlying cloud service provider takes to provision some services, so to not keep you waiting too long, we interrupt the process, and allow you to create a new deployment. So, you don’t need to worry, we won’t count this deployment. Please create a new one.