The management of data resources and the int index limitation of computers is always a great bottleneck in big data. For example, ARIMA is one of the most important statistical models. For the int index limitation, in 32-bit computers, the maximum seasonality supported by ARIMA is 350, or ARIMA will run out of memory; this bottleneck cannot be overcome by increasing the computer's memory. In this paper, a new fuzzy system model – a type-2 fuzzy event parallel computing system is proposed with novel type-reduction computational optimisation technique and type-2 fuzzy probabilistic parallel computing, to overcome the bottleneck. Being different from conventional fuzzy logic systems whose inference engine are based on IF-THEN rules, the new system incorporates statistical inference with type-2 fuzzy events. The system extends ARIMA to provide unlimited maximum seasonality support, at the same time, it retains the benefits of ARIMA in accurate computing and the specialties of fuzzy logic in uncertainty modelling. The system is applied in real world long period uncertain data-intensive seasonal time series – Wireless Soft-Switch CAPS (Call Attempt Per Second) forecasting. As a series of concepts, algorithms, experiments and results prove, a type-2 fuzzy event parallel computing system is viable in practice, which realises fuzzy system models evolve from the models based on IF-THEN rules to the models incorporating statistical inference with type-2 fuzzy events.