Data being used in real-time systems must be up-to-date to produce correct results. The use of outdated data can have catastrophic consequences since calculated control signals are based on stale data. Two distinct methods to update data exist: (i) dedicated tasks (DT) update data items, and (ii) on-demand (OD) updating being a conditioned part of the execution flow of tasks. On-demand updating has not been studied in terms of CPU utilization analysis for real-time systems. This paper studies on-demand updating in terms of (i) imposed workload and compares the workload to Deferrable Scheduling (DS), and (ii) analytical formula for estimating workload to be used in CPU utilization based schedulability tests. It is found that on-demand updating uses less workload for updates compared to DS, which suggests on-demand updating should be used for resource constrained systems. However, using on-demand updating makes the execution times of updates unpredictable, which currently gives two possibilities (i) be pessimistic and assume all updates always execute or (ii) be less pessimistic but estimate the times between executions of updates. This paper devises a formula for such estimates and compares their result to approach (i). Evaluations show the formula can be useful for soft real-time systems.