The idea of Industry 4.0 includes the concept of Industrial Internet of Things (IIoT) that is the possibility for industrial devices to have Internet connection and share data. Huge amount of data are stored and analyzed in the Cloud to extract meaningful information to be sold as 'services'. Today, many Industry 4.0 scenarios do not require a short latency between data collection and output reaction, but it is expected that short latency services would be seen by the market as a distinctive quality. This paper deals with the estimation of latency in transferring data from the field (where the production takes place) to the Cloud and then back to field. Since IIoT natively refers to worldwide applications, the paper analyzes some cases where interacting nodes are deployed in different continents. The experimental results show that simple solutions based on widely accepted lightweight protocols (e.g. MQTT) and inexpensive industrial grade IoT devices are feasible. From the performance point of view, when using free access Cloud servers, they can achieve round trip latency down to 300 ms with standard deviation of about 20 ms over one-week observation time.