VaultSpeed Automation Unit (VAU)
Based upon the subscription type, customers are entitled to a certain amount of VaultSpeed Automation Units (VAUs) per subscription (year). On this page, we define what a VAU is and how different processes consume them.
A VAU is a unit used by VaultSpeed which tracks the intensity of resource usage by a customer. This is one of the most important factors to consider when choosing a subscription type, further in this document we will provide a way to estimate this value based on the number of source objects and the frequency of change.
The main consumer of VAUs is jobs counted when generating code. How these jobs are calculated is explained later in this document.
At the start of each subscription, we will count some VAUs based on the size of your Data Vaults. This count is comparable to a full generation.
The processes which generate VAUs are:
Code generation jobs. Each job counts as 1 VAU. See the next section for how these are defined and calculated.
Metadata exports, the counting algorithm is explained in https://vaultspeed.atlassian.net/wiki/spaces/AF/pages/1897988236.
Source copies, each source object copied counts as 1 VAU.
What is a job?
To understand the definition of a job in VaultSpeed, you must understand how Data Vault works. A source model with 3 tables and 3 foreign keys will result in a data vault model that consists of 3 HUB’s, 3 or more SAT’s on HUB, 3 Links and 3 or more SAT’s on LINK. Loading this model will require ELT flows for each one of these Data Vault objects.
The cost of generating a Data Vault in VaultSpeed is directly related to the complexity of the source model. This makes sense: a large, complex source model with lots of relationships between tables will result in a large and complex data vault model and therefore requires more compute both in DDL, ETL and Workflow generation. So the number of jobs in VaultSpeed is equal to the number of flows needed to load all elements needed in the target model (raw data vault model, business vault model…). One single element includes its DDL and ETL generation.
1 VaultSpeed job = 1 flow to load an element (HUB,SAT,LINK,…) in the target model
Total job count
The total job count is calculated based on the number of jobs. We make a few distinctions in the calculation:
There is a difference between existing logic that needs to be regenerated (due to parameter changes, a new VaultSpeed version,…) and the generation of new logic (new source tables, attributes, relations…).
A large Data Vault will require higher metadata storage than a relatively small Data Vault.
Job counter = # jobs with NEW logic + ¼ x # jobs with EXISTING logic + # DV objects
Each part of this formula plays a vital role in the determination of a customers job usage:
Whenever new logic is added to the data warehouse, we will add a full job to the count. This will typically be the case in any delta generation because a delta only generates the changes to your model. And also when the first initial setup is generated.
When existing logic is regenerated, we only take into account a quarter of a job. This will typically happen whenever you regenerate an existing release.
And at the start of each license year, we account for data vault size and metadata storage costs by adding the number of objects inside the data warehouse. This metric also enables us to consider the amount of computing needed when comparing between releases for delta generation.
Your job counter will be reset automatically at the start of each license year (upgrading your licence sets the start of a new licence year)
Example:
Job count estimate
To estimate the number of jobs needed, you can use the following estimation guide rule. Be advised that for selecting the best license fit, you also need to consider the number of concurrent users and additional support.
First, we estimate the number of Data Vault objects. This is based on averages we see across our customer base.
DV^{E} = Estimated # DV objects = 5,5 x #source tables
1 HUB + 1,2 SATs + 1,6 LINKs + 1,7 SATLINKs
You might notice that the number of satellites is slightly higher than the respective number of HUBs and Links. This is due to satellite splits and other data vault model decisions.
Yearly estimated # of jobs needed = 2 * DV^{E} + DV^{E} * # source releases * %change
The yearly estimated number of jobs depends on the number of objects in your data vault, the number of source releases you can expect each year, and the number of objects changed in the source model.
This indication will give you a ballpark figure of what to expect based on the number of source tables you want to integrate into your data warehouse.
Example:
%changed objects  20% 
 # source releases  12 (monthly) 

# Source tables  DV^{E}  Bronze  Silver (10.000 jobs)  Gold (60.000 jobs)  Platinum (100.000 jobs) 
<40  220  ✓




<400  2200 
 ✓ 


<2450  13475 

 ✓ 

<4000  22000 


 ✓ 
VAU Counter Dashboard
The VaultSpeed frontend environment gives you a clear view of how many VAUs you have left in your subscription. This overview can be found on the dashboard page. We also have a warning limit when you reach 80% of the limit for your subscription.