Analyze and Understand Licensing

Analyze and understand licensing costs in your workload

We will analyze the CUR file for any software that includes licensing. First we will see what columns could give us the information we need. Then we look for the amount we are spending on items containing licenses, and how much we are spending on the actual licenses by comparing to a similar non-licensed option.

We take this information and decide if the effort required to make the change will be less than what we save, and if the required functionality is still met.

Analyze CUR columns

The first step is to understand what information we may have that can show us costs associated with licenses.

1. Go to the Athena Console:

2. Select the costmaster database, you should see the before and after tables

3. We want to see all the columns available, and some sample data, so paste the following command and click Run query:

    SELECT * FROM "costmaster"."before" limit 10;
   

4. You can see each column and 10 rows of data below. There are over 100 columns - take a minute or two and slowly scroll through them and view the data.

5. The supplied datasource has multiple workloads in it, we want to ensure we are only looking at costs for our ordering workload. There is a column at the far right resource_tags_user_application, this contains the tags we put on our resources. Lets limit our search to costs tagged with ordering. Run the query:

    SELECT * FROM "costmaster"."before" 
    where resource_tags_user_application like 'ordering'
    limit 30
   

6. Now we will look for EC2 Operating System licenses. Look through the data and see if you can find a column which indicates the operating system used.

7. There are multiple columns we can use, look at the columns line_item_line_item_description and product_operating_system. Note: columns are close to alphabetical order.

8. We will focus on line_item_line_item_description, you can see Linux, Windows and RHEL. We will get the costs of RHEL licensing and compare it to a similar operating system, AWS Linux. To get a sample of the RHEL costs, paste the following query and click Run query:

    SELECT * FROM "costmaster"."before"
    where resource_tags_user_application like 'ordering' and line_item_line_item_description like '%RHEL%'
    limit 30
   

9. We will now get the costs for each hour for running RHEL. We dont need 100+ columns, so we’ll focus on just the following columns: line_item_usage_start_date, linte_item_line_item_description, line_item_unblended_cost. We will sum the unblended_cost column, and group it by usage_start_date, this will give us cost per hour. The order by line orders the output by date. Run the following query:

    SELECT line_item_usage_start_date, line_item_line_item_description, sum(line_item_unblended_cost) as cost FROM "costmaster"."before"
    where resource_tags_user_application like 'ordering' and line_item_line_item_description like '%RHEL%'
    group by line_item_usage_start_date, line_item_line_item_description
    order by line_item_usage_start_date asc
   

10. You can see that the cost per hour of running the RHEL instances was 0.8128000000000001 per hour. To get the pricing for a non-licensed operating system we will need the instance size and location. We will use the line_item_usage_type and line_item_availability_zone columns to get this, run the following query:

    SELECT line_item_usage_type, line_item_availability_zone, sum(line_item_unblended_cost) as cost FROM "costmaster"."before"
    where resource_tags_user_application like 'ordering' and line_item_line_item_description like '%RHEL%'
    group by line_item_usage_type, line_item_availability_zone
    limit 30
    

11. Look at the columns: line_item_usage_type, line_item_availability_zone, you will see the values BoxUsage:t3.medium and us-east-1a/b:

12. We see that we are running t3.medium RHEL instances in us-east-1. Open the pricing pages in a new tab: https://aws.amazon.com/ec2/pricing/on-demand/

13. Ensure the Linux tab is selected, and the region is US East (N. Virginia). The pricing for a t3.medium with AWS Linux is: $0.0416 per hour. Now click the RHEL tab:

14. The pricing will change, the price for RHEL is: $0.1016.

15. Lets see exactly how many hours of usage we were running in total (line_item_usage_amount), and the price. In the Athena console paste the following query:

    SELECT line_item_line_item_description, sum(line_item_usage_amount) usage_hours, sum(line_item_unblended_cost) as cost FROM "costmaster"."before"
    where resource_tags_user_application like 'ordering' and line_item_line_item_description like '%RHEL%'
    group by line_item_line_item_description
    

16. You can see we consumed 176 instance hours of licensed RHEL ($0.1016 an hour), for a total cost of $17.8815999….

17. Our sample CUR is less than a month, to get monthly costs - lets see how many instances we are running with RHEL. We will get the resource IDs and how long they ran for with the following query. In the Athena Console paste the following:

    SELECT distinct line_item_resource_id, sum(line_item_usage_amount) as Hours_ran  FROM "costmaster"."before"
    where resource_tags_user_application like 'ordering' and line_item_line_item_description like '%RHEL%'
    group by line_item_resource_id
    

18. We are running 8 instances of RHEL at a cost of $0.1016 per hour. If we changed these to AWS Linux at a cost of $0.0416 per hour, we would save $0.06 each hour, per instance.

19. 8 Instances at saving of $0.06 per hour, for an average of 730 hours each month, is a monthly saving of $350.4.

Understand the costs of running licensed software in your workload

There can be associated costs with running licensed software. Additional resources may be mandated by the software in addition to a base configuration. We will continue our example by looking at operating systems.

We will now see if additional resources are required to run RHEL compared to AWS Linux, and the cost of these additional resources.

1. Go into the EC2 console, and click Launch Instance, then Launch instance

2. Select the Amazon Linux 2 AMI

3. Select a t3.medium instance size, and click Next: Configure Instance Details

4. Click Next: Add Storage

5. You can see that the default configuration is for 8GiB GP2 of EBS storage. Click Cancel.

6. From the EC2 Console click Launch Instance, then Launch instance

7. Select the Red Hat Enterprise Linux AMI:

8. Select a t3.medium instance and click Next: Configure Instance Details:

9. Click Next: Add Storage

10. You will see that there is 10GiB GP2 of storage required for RHEL as a default. Click Cancel.

11. Confirm how much storage we are using and the cost:

     SELECT line_item_usage_start_date, line_item_line_item_description, sum(line_item_unblended_cost) as cost, sum(line_item_usage_amount) as amount FROM "costmaster"."before"
    where resource_tags_user_application like 'ordering' and line_item_usage_type like '%EBS%' and resource_tags_user_tier like 'front'
    group by line_item_usage_start_date, line_item_line_item_description
    order by line_item_usage_start_date asc
    limit 5
    

12. Each hour we are consuming 0.111111 for a cost of $0.011111.

13. You need an additional 2Gb of GP2 storage per instance. Go to the EBS pricing page here: https://aws.amazon.com/ebs/pricing/

14. The price in US East(N. Virginia) is $0.1 per GB-month for SSD gp2 volumes. We will now calculate the storage savings across our workload.

15. Lets see how many instances we are running with RHEL. We will get the resource IDs and how long they ran for with the following query. In the Athena Console paste the following:

    SELECT distinct line_item_resource_id, sum(line_item_usage_amount) as Hours_ran  FROM "costmaster"."before"
    where resource_tags_user_application like 'ordering' and line_item_line_item_description like '%RHEL%'
    group by line_item_resource_id
    

16. We are running 8 instances, which each require an additional 2GB GP2. This is 16GB-month of storage (at $0.10 per GB-month), which is an additional cost of $1.60 each month on our workload.

17. We have total additional monthly costs due to licensing of $350.4 (licenses) + $1.60 (storage) = $352

Simulate the change and validate

We have performed the change in our environment and created new Cost and Usage Reports. We will now analyze the after table which contains these changes.

1. Lets see the amount and cost of our new AWS Linux instances, and the price. In the Athena console paste the following query:

    SELECT distinct line_item_resource_id, line_item_line_item_description,  sum(line_item_usage_amount) as Hours_ran,  sum(line_item_unblended_cost) as cost   FROM "costmaster"."after"
    where resource_tags_user_application like 'ordering' and line_item_line_item_description like '%t3.medium%' and resource_tags_user_tier like 'front'
    group by line_item_resource_id,  line_item_line_item_description
   

2. You can see we ran 8 Linux instances in a sample of 23hrs at a cost of $0.09568 for each instance.

3. The total cost for 8 instances over 730 hours ($0.9568 / 23 * 730 * 8), would be $242.944. Our original cost was $593.344, and we predicted a saving of $350.4, our final cost of $242.944 verifies this prediction.

4. Lets analyze the storage with the following query:

    SELECT line_item_usage_start_date, line_item_line_item_description, sum(line_item_unblended_cost) as cost, sum(line_item_usage_amount) as amount FROM "costmaster"."after"
    where resource_tags_user_application like 'ordering' and line_item_usage_type like '%EBS%' and resource_tags_user_tier like 'front'
    group by line_item_usage_start_date, line_item_line_item_description
    order by line_item_usage_start_date asc
    limit 5
   

5. You can see there was an hourly amount of 0.088888 and an hourly cost of $0.008888, which is 8/10 of the original amount of 0.111111 and cost of 0.011111, which validates changing to 8Gb volumes from 10Gb volumes. This validates our $1.60 saving prediction.