How is link utilization calculated in the Capacity Planner?
The Capacity Planner handles a number of special situations that complicate its algorithms, but fundamentally, link utilization is calculated as the load on the link divided by the capacity of the link. For example, 512 Kbps of data on a 1024 Kbps link gives 512 / 1024 = 0.50, or 50% utilization.
The capacity of the link just a specified value. All types of links have one or more default capacity values, depending on the technology, and the user can change, remove or add capacity values. Part of the network optimization that the Capacity Planner performs is choosing between alternate capacities for each link to better match the load on that link.
The load on a link is based on all types of traffic specified in the network, such as IERs, applications, flows and background loads. The Capacity Planner selects the traffic that is active at a particular time, routes the messages on the network and counts the amount of traffic that each link would carry at that time. That amount of traffic is the load. The following example illustrates how the Capacity Planner calculates the load.
Assume that the network contains two Ethernet workstations connected by 10BaseT, and that there is one IER between the workstations. The IER has size 32,000,000 bytes and fires just once.
First, note that the Capacity Planner actually evaluates the network during a time period, not at an instant of time. By default, the time period is the first hour of simulated time. In this example, all of the traffic (one IER) is active during that time period. (If the IER's Start Time were changed to 3700 seconds, it would not fire in the first hour and would be ignored.)
The Capacity Planner then converts the size of the IER to bits and divides that value by the duration of the time period. This converts the IER into a constant capacity requirement across the time period. Since there is only one IER, this is the total "load" on the link.
(32,000,000 bytes x 8 bits/byte) / 3600 secs ~= 71,111.11 bps ~= 71.11 Kbps
The link is 10BaseT, so the utilization is calculated as:
71.11 Kbps / 10,000 Kbps = 0.007111 or 0.71%
In these calculations, it is obvious that changing the size of the IER will directly affect the utilization. Likewise, adding a second, identical IER or allowing the original IER to fire repeatedly will increase the load and change the utilization.
In a more complicated network, other behaviors can also affect the traffic, such as data and circuit routing, but in simple cases, the results can be easily traced to these calculations.
By design, the Capacity Planner does not consider the many of the details of data transmission, including frame size and window size. The Capacity Planner calculations purposely omit detail to provide results more quickly. On the other hand, Discrete Event Simulation does consider all of those details, and usually requires more time to provide results. The two tools allow the user to trade-off speed and fidelity. A common work-flow is to use the Capacity Planner to "sanity check" a network before running a Discrete Event Simulation. However, the Capacity Planner's accuracy is sufficient for many users and they use it instead of Discrete Event Simulations.
The speed of the Capacity Planner also makes Capacity Optimization feasible. The Optimization feature iteratively searches for better network capacity settings and requires repeated link utilization calculations. Discrete Event Simulation would generally be too time-consuming for this process.