Abstract: The Gamer's Non-public Community (GPN) is a client/server expertise created by WTFast for making the network performance of on-line video games sooner and more reliable. GPN s use middle-mile servers and proprietary algorithms to better connect online video-sport gamers to their recreation's servers throughout a wide-area community. Online games are an enormous leisure market and community latency is a key aspect of a participant's competitive edge. This market means many alternative approaches to network architecture are carried out by totally different competing companies and that these architectures are constantly evolving. Ensuring MINECRAFT TOWNY SERVERS between a shopper of WTFast and the web sport they wish to play is thus an incredibly difficult problem to automate. Utilizing machine learning, we analyzed historical network information from GPN connections to explore the feasibility of community latency prediction which is a key part of optimization. Our subsequent step will be to collect dwell data (together with shopper/server load, packet and port data and particular recreation state data) from GPN Minecraft servers and bots. We'll use this information in a Reinforcement Learning mannequin along with predictions about latency to change the clients' and servers' configurations for optimum community performance. These investigations and experiments will enhance the standard of service and reliability of GPN programs.