Serverless or Managed

[Frank]

We have about 12k write/s and 1.5TB/mo compressed S3 data.

How can we choose between Serverless vs managed? And what will be good way to project the cost?

In serverless, how the machine and hours scaled or scheduled based on the load?

If there is a long query, can it be cancelled?

[Prabakar · Accepted solution]

Hi @Frank Zhang​ 

How can we choose between Serverless vs managed? And what will be good way to project the cost?

-- Once you enable the serverless feature on your workspace, by default the new warehouse will be created with a serverless option. If you dont want to use serverless for the workload, you can disable serverless from the configuration page.

 

In serverless, how the machine and hours scaled or scheduled based on the load?

-- Serverless and managed works the same way. The difference is Serverless is managed by Databricks and we have the instances available so the cluster startup time will be much faster. During public preview, Serverless compute will launch VMs for clusters in under 15 secs, whereas when we are GA, its expected to reduce to under 10secs.

If there is a long query, can it be canceled?

-- Yes it can be canceled.

[Kaniz]

Hi @Frank Zhang​ , We haven't heard from you on the last response from @Prabakar Ammeappin​​, and I was checking back to see if his suggestions helped you.

Or else, If you have any solution, please share it with the community as it can be helpful to others.

Also, Please don't forget to click on the "Select As Best" button whenever the information provided helps resolve your question.