Business driver map
A business driver map is a key component of capacity planning. It represents the association between business drivers and the system resources that they are consuming.
For example, a simple web application uses two systems: an application server (AS) and a database (DB). A large amount of application resources are used to resolve page requests issued by web clients. The main business driver is represented by a metric – page requests per hour. Each request is processed by the application server and potentially queries the database.
There is a Send Message page on the application server through which messages are sent to communications middleware. Therefore, each message sent (each visit made to that page) consumes additional application server resources in terms of CPU usage, but does not require database access.
Based on this information, an initial association between business drivers and consumed resources is built, as shown in the following table:
AS-CPU Utilization %
AS-Disk Utilization %
DB-CPU Utilization %
DB-Disk Utilization %
These associations represent the first step to the creation of a business driver map.
Note the following key points:
- Associations are always made between metrics and not between systems and business drivers. A specific business driver object can consume a particular set of resources of a system (for example, the CPU) without affecting the others.
- Each business driver map is associated with a specific analysis. For each performance versus load analysis (PLA) that you create inside an application, you can use a different business driver map to show associations. In this way, you can build and validate different models of your service.
When working with shared systems, the business driver of an application (for example, application B) might use the system at the same time as the business driver of another application (application A). In this scenario, application A is said to have external business drivers. While entering data into the business driver map, determine whether the application being analyzed contains shared systems.
The following figure shows an example of how a business driver map for a performance versus load analysis appears in BMC TrueSight Capacity Optimization.
Business driver map of a performance versus load analysis – Associations tab
The main objectives of this business driver map are:
- Plotting the analysis
- Estimating the service demands for the Working with queuing network models
- Estimating the cost of business drivers
After building initial associations, the business driver map analysis helps you understand system behavior by means of the following automatically calculated indicators:
Understanding the map
The values of these calculated indicators can help validate the business driver map and understand whether the models based on that map are reliable for predictions or not.
To better understand the analysis, focus on the following:
- Each business driver or correlation map row is a separate analysis case.
- If a chosen row has a single association (indicated by an X), the analysis for this resource or business driver is univariate (or one-dimensional, 1-D).
- If a chosen row has multiple associations (indicated by an X), the analysis for this resource or business driver is multivariate (or N-dimensional, N-D).
A univariate analysis, considers each business driver separately and tries to explain the entire utilization of the system with each business driver. This method can be used if the business drivers are highly correlated, for example, the number of active transactions and the number of active users on a system. This type of analysis is generically more suited for pure correlation analyses like Performance vs Performance and Load vs Load.
In the case of the business driver map, you have a system metric (for example, U) and a resource metric (for example, W). Crossing their values on a scatter plot in the U/W plan, you obtain a graph that is called performance versus load analysis. If U is highly correlated with W, then:
- The points are disposed in a more or less wide cloud around a line.
- The solution of the univariate analysis is the linear regression of the cloud, meaning, a line having an intercept that we can call residual utilization (U0) and an angular coefficient that is called service demand (D) in capacity planning. The line represents the equation
U = U0 + D * Wand it is the service demand line.
- The linear models based on the calculated service demand can be used for predictions. For example, to calculate the U value for a given W.
However, if the U/W correlation value is low, the cloud in the PLA plot will be more scattered and there will likely be no evidence of a service demand line.
Using a multivariate analysis, the utilization on the system is explained by the sum of all the business drivers in the map. This is the ideal method to select to perform a queueing network or extrapolation model to forecast how much the utilization will be based on the growth of the business driver.
Multivariate analysis is more complex to visualize because there is no graphic representation of the issue. To understand the solution of a multivariate analysis, imagine a N-dimensional surface that represents the linear regression of the N-dimensional U/W cloud.