作者: Javier Alonso , Inigo Goiri , Jordi Guitart , Ricard Gavalda , Jordi Torres
关键词:
摘要: Nowadays, virtualized platforms have become the most popular option to deploy complex enough services. The reason is that virtualization allows resource providers increase utilization. Deployed services are expected be always available, but these long-running especially sensitive suffer from software aging phenomenon. This term refers an accumulation of errors, which usually causes exhaustion, and eventually makes service hang/crash. To counteract this phenomenon, a preventive approach fault management, called rejuvenation has been proposed. In paper, we propose framework provides transparent predictive web on platforms, achieving high levels availability. exploit provider resources, also seeks maximize number running simultaneously platform, while guaranteeing resources needed by each service.