This lightning talk provides an in-depth overview of Envoy's memory allocation management system, focusing on its integration with Google's tcmalloc allocator. Learn how Envoy handles memory allocations, explore the on-demand memory releasing techniques and understand why they sometimes behave unpredictably. The presentation also covers future plans for supporting additional allocator types that could improve memory management for specific deployment scenarios. Delivered by Kateryna Nezdolii from Docker, this 13-minute technical session offers valuable insights for cloud native practitioners working with Envoy proxy.