Gecko: a contention-oblivious design for cloud storage
Ji-Yong Shin, Mahesh Kumarr Balakrishnan, Lakshmi Ganesh, Tudor Marian, Hakim Weatherspoon · USENIX conference on Hot topics in storage and file systems · 2012
Disk contention is a fact of life in modern data centers, with multiple applications sharing the storage resources of a single physical machine. Log-structured storage designs are ideally suited for such high-contention settings, but historically have suffered from performance problems due to cleaning overheads. In this paper, we introduce Gecko, a novel design for storage arrays where a single log structure is distributed across a chain of drives, physically separating the tail of the log (where writes occur) from its body. This design provides the benefits of logging - fast, sequential writes for any number of contending applications - while eliminating the disruptive effect of log cleaning activity on application I/O.