: Experimental results using the DeathStarBench benchmark showed that TPRAM can save at least 40.58% of CPU and 15.84% of memory resources while maintaining end-to-end Quality of Service (QoS). Accessing the Paper
: It uses a Transformer-based attention mechanism to build a performance prediction model for microservice nodes on a system's "critical path".
The file name is a shorthand for the framework (Transformer-based Prediction and Resource Adaption Method) and likely one of its primary authors or a related contributor, such as Yang Chen or Hongyan Xia (whose research is often associated with these models). Paper Summary: TPRAM TpRam-Kelly.7z
: It employs Deep Deterministic Policy Gradient (DDPG) , a reinforcement learning technique, to dynamically adjust CPU, memory, and I/O disk allocation based on real-time requirements.
The paper addresses the difficulty of optimizing resource allocation in cloud-native environments where microservices have complex dependencies. Paper Summary: TPRAM : It employs Deep Deterministic
: A preprint or abstract of the work is hosted on ResearchGate .
You can find the full text or official citation through these platforms: You can find the full text or official
The file refers to the research paper titled " Transformer-based performance prediction and proactive resource allocation for cloud-native microservices ," published in Cluster Computing in August 2025.
