Azure Data Lake Analytics: More Compute and More Control


Customers have been telling us that they want access to more computational horsepower for running U-SQL jobs. Today, we are announcing that we are enabling:

  • More parallelism available to each job, raising it from 60 to 120 units.
  • Increasing the maximum number of jobs that can run concurrently, raising it from 3 jobs to 20 jobs.

Before today, if you wanted to change these limits you’d have to contact us via product support, starting today you can directly change these settings in the Azure Portal.

 

 

Frequently Asked Questions:

How can I change the maximum number of concurrent U-SQL Jobs?

This number is set to 10 by default. This means that each account can have at most 10 concurrent jobs running at the same time. You can increase this setting up to 20. To change this, go to your ADLA account > Properties > Concurrent Jobs

 

What happens if I use all my parallelism in 1 job and then submit another job?

The next submitted job will not start immediately. It will be queued until there is enough parallelism available.

 

How can I change the max amount of parallelism for my account?

This number is set to 120 by default. Meaning each ADLA account can use up to max of 120 analytics unit at any given time but admins can decrease this number for their account in order to control the billing costs. To change this, go to your ADLA account > Properties > Parallelism

 

Does this change affect the cost of using an ADL Analytics account?

The pricing for Analytics Units did not change.  However, you can now use more Analytics Units which means you can spend more money.

 

Why do I still have the old limits in my account?

All trial subscriptions are still bound to the old limits. Please check your subscription type to make sure it is not “Trial”.

 

What should I do if I need more than 120 parallelism or 20 concurrent jobs?

Please open a support request in Azure Portal. We will review your request and try to accommodate your business needs.

 


Comments (0)

Skip to main content