SeRW: Adaptively Separating Read and Write upon SSDs of Hybrid Storage Server in Clouds

Nowadays, cloud providers embrace hybrid storage servers to reap both high IO performance of solid-state drives (SSDs) and low-cost of hard disk drives (HDDs). These hybrid storage servers generally employ SSDs as primary storage directly serving requests from front-end applications while using HDDs as the secondary storage to provide sufficient storage capacity. From an actual trace analyzing on the production of Alibaba clouds, we found that writes mixed with mid/high intensive reads upon SSDs dramatically increases read-latency, especially for tail latency. These long read latencies are primarily caused by (1) write-induced-blocking and (2) write-induced-garbage-collection (GC). To fully exploit the IO potential of underutilized HDDs within hybrid storage servers, we present a SeRW scheduling approach that adaptively steers some SSD-writes to idle HDDs in running time. SeRW relieves the write-blocking read delay on SSDs at mid/high load. SeRW also reduces the amount of data written into SSDs, thus decreasing the GC frequency. We evaluate SeRW through replaying the production workload. The experimental result shows that SeRW decreases the average, 99th, 99.9th, 99.99th-percentile latencies of reads by up to 2.07x, 1.48x, 4.29x, and 4.24x, respectively, while reducing the amount of data written to SSDs by up to 37.5%. SeRW notably significantly improves both the overall system performance and SSD endurance without sacrificing the write latency.