ETRI-Knowledge Sharing Plaform



논문 검색
구분 SCI
연도 ~ 키워드


학술대회 Predictive Mechanism for Server Health Check: Network Packet-Flow Learning Approach
Cited 0 time in scopus Download 0 time Share share facebook twitter linkedin kakaostory
손승철, 고석갑, 이형옥, 이병탁
International Conference on Information and Communication Technology Convergence (ICTC) 2022, pp.2095-2097
22ZK1100, 호남권 지역산업 기반 ICT 융합기술 고도화 지원사업, 강현서
Large-scale Internet services are integrated into physical server clusters and virtual microservices. For smooth service, failures of individual microservices and server clusters constituting integrated services should be checked from time to time. Heartbeat packets are a typical method, but they bring network overhead. In this paper, we propose a method to passively inspect the health of individual microservices and individual servers through machine learning on network packets. Proposed techniques allow us to predict the health status of individual microservices and individual servers without incurring additional network overhead. Experimental results show that the ROC-AUC of the proposed technique is close to 0.92 and is reasonable except for very few false positive results.
KSP 제안 키워드
False positive results, Integrated Services(InterServ), Internet service, Learning approach, Network overhead, Packet-flow, Predictive mechanism, Server cluster, Server health, health status, large-scale