Human-in-the-Loop Automatic Data Migration for a Large Research Computing Data Center

Fang (Cherry) Liu, Michael D. Weiner, Kevin Manalo, Aaron Jezghani, Christopher J. Blanton,Christopher Stone, Kenneth Suda, Nuyun Zhang, Dan Zhou,Mehmet Belgin, Semir Sarajlic,Ruben Lara

2021 INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND COMPUTATIONAL INTELLIGENCE (CSCI 2021)(2021)

引用 3|浏览10
暂无评分
摘要
Most HPC centers face a lack of expertise of data center migrations, as it's a rare event that only a small portion of HPC professionals experience more than once in their entire professional careers. This paper presents how the Georgia Institute of Technology (Georgia Tech) Partnership for an Advanced Computing Environment (PACE) team employed automation to migrate research computing data from the old Rich computing center (Rich) to the new Coda data center (Coda) in 2020. PACE successfully migrated 1844 TB of data for 3550 users without loss of user data. PACE implemented a `human-in-the-loop' automatic workflow to facilitate the migration, interleaving automated scripts with human-driven reviews, significantly reducing staff time commitment while ensuring the integrity and accuracy of data migrations. PACE deployed a cached data movement strategy which reduced the migration downtime significantly. We share our one-year migration journey for the benefit of the HPC community.
更多
查看译文
关键词
data center,data migration,automatic process
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要