Live migration improvements by related dirty memory prediction in cloud computing. (15th July 2017)

• Record Type: Journal Article
• Authors: Wu, Tin-Yu; Guizani, Nadra; Huang, Jhih-Siang
• Is Part Of: Journal of network and computer applications. Volume 90(2017)
• Page Start: 83
• Page End: 89
• Publication Date: 2017-07-15
• Subjects: Pre-copy migration -- Memory prediction scheme -- Writable Working Set (WWS); Microcomputers -- Periodicals; Computer networks -- Periodicals; Application software -- Periodicals; Micro-ordinateurs -- Périodiques; Réseaux d'ordinateurs -- Périodiques; Logiciels d'application -- Périodiques; Application software; Computer networks; Microcomputers; Periodicals; 004.05; 004
• DOI: 10.1016/j.jnca.2017.03.011
• Languages: English
• ISSNs: 1084-8045

Abstract

Abstract: To solve the problem of too long pre-copy migration time, pre-copy migration methods, including post-copy, time-series based pre-copy method, etc., are therefore proposed. Pre-copy methods can decrease the total migration time but may result in the longer downtime. There are three phases of memory migration: (1) Push Phase, (2) Stop-and-copy Phase, and (3) Pull Phase, and there are two key performance indices of live migration: downtime and total migration time. For live migration enhancement, we propose a memory prediction mechanism, which can choose the pages to migrate in the iterative pre-copy phase or in the stop-and-copy phase based on related dirty rate. According to the features of memory pages, our proposed mechanism is able to decide the best time to perform memory migration so as to decrease not only unneeded migrations but also the total migration time (Hines et al., 2009).