Analysis Of And Optimization For Write Workloads Of Cloud Storage Nodes

Nowadays large commercial cloud providers,such as Alibaba Cloud,need to provide users with high-quality services at the lowest cost.Therefore,these cloud datacenters widely deploy hybrid storage nodes consisting of Solid State Drives(SSDs)and Hard Disk Drives(HDDs),reaping their respective benefits: performance from SSD and capacity from HDD.These hybrid storage nodes usually write the incoming data to the SSD first to ensure a low write latency,and then the background threads write it to the corresponding HDD in batches.This write mode is called SSD Write Back(SWB)mode.When comprehensively analyzing real production workloads from Pangu,a large-scale storage platform underlying the Alibaba cloud,we find that(1)there exist many write dominated(the propotion of write exceeds 77%)hybrid storage nodes(WSNs);however,(2)under the SWB mode,the SSDs of these WSNs suffer from severely high write intensity and long tail latency.To address these unique observed problems of WSNs,we present SSD Write Redirect(SWR),a runtime IO scheduling mechanism.SWR judiciously and selectively forwards some or all SSD-writes to HDDs,adapting to runtime conditions.By effectively offloading the right amount of write IOs from overburdened SSDs and strategically forwarding them to underutilized HDDs,SWR can greatly improve the write performance of WSNs and SSD endurance.Our trace-driven evaluation of SWR,through replaying production workload traces collected from the Alibaba cloud in our cloud testbed,shows that SWR decreases the average and 99th-percentile latencies of SSD-writes by up to 13% and 47% respectively,notably improving system performance.Meanwhile the amount of data written to SSDs is reduced by up to 70%,signifcantly improving SSD lifetime.