Instantaneous observability is a method used here to watch a system whose internal states vary very fast. It is a system property which allows us to estimate the system internal states. A discrete computing process which is time-varying can be represented by a mathematical model. This model has linear, discrete, stochastic and time-varying equations which contain matrices and vectors whose elements are deterministic functions of discrete time k. This computing process is performed during a period of time called latency. The process is performed in four steps in sequence, the system states: new process, ready-queue, CPU running, process end. The system propertyinstantaneous observabilitydepends on the pair of matrices {A(k), C(k)} and regards the possibility to estimate the system internal states when the system state equations are known. The problem is the system states are inside and they are not always accessible directly. In this paper we propose a method to determine: the instantaneous observability matrices at discrete time k, the system state estimation and the system latency when the system output measurements are known. We will show when instantaneous observability property comes true, the system instantaneous internal states and latency can be estimated. This is an advantage compared to usual observability method based on static scenarios. The potential application of the results is a prediction of data traffic-jam on a computer process. In a broader perspective, the instantaneous observability method can be applied on identification of pathology, weather forecast, navigation, tracking, stock market and many other areas Ver menos