Hi,
I am working on a embedded application that needs to occasionally connect to a server. The overall response time for each communication is very critical. I have been experimenting with a simple example in a linux VM (instead of the ultimate embedded system) and also Amazon cloud (AWS). The latency for my local VM is relatively large (mean 0.6) while for AWS I got much less latency (0.3 and less). Now my question is :
1- Is there a way to give this specific server IP high priority so the latency drops?
2- Is there a way to improve the "speed" for application that involves streaming ( relatively short stream of data)?

thanks

Only by implementing some kind of QoS in your network eg 802.11p for layer 2 prioritization and/or differentiated services for layer 3. Outside my scope of knowledge, so I'll leave it to you to research further.

https://en.wikipedia.org/wiki/IEEE_P802.1p
https://en.wikipedia.org/wiki/Differentiated_services

The latency from where to where?

The actual *Internet* pipes, controlled by other companies, aren't going to care about how important you think your packets are

See sysctl tweaks, e.g. tcp_slow_start_after_idle