Analytic modeling of SSD write performance

• Authors: Peter Desnoyers
• Venue: Proceedings of the 5th Annual International Systems and Storage Conference
• Year: 2012

Solid state drives (SSDs) update data by writing a new copy, rather than overwriting old data, causing prior copies of the same data to be invalidated. These writes are performed in units of pages, while space is reclaimed in units of multi-page erase blocks, necessitating copying of any remaining valid pages in the block before reclamation. The efficiency of this cleaning process greatly affects performance under random workloads; in particular, in SSDs the write bottleneck is typically internal media throughput, and write amplification due to additional internal copying directly reduces application throughput. We present the first precise closed-form solution for write amplification under greedy cleaning for uniformly distributed random traffic, and validate its accuracy via simulation. In addition we also present the first models which predict performance degradation for both LRU cleaning and greedy cleaning under simple non-uniform traffic conditions; simulation results show the first model to be exact and the second to be accurate within 2%. We extend the LRU model to arbitrary combinations of random traffic, and demonstrate its use in predicting cleaning performance for real-world workloads. The utility of these analytic models lies in their amenability to optimization techniques not feasible in simulation. We examine the strategy of separating "hot" and "cold" data, showing that for our traffic model, such separation eliminates any loss in performance due to non-uniform traffic. We show how a system which separates hot and cold data may shift free space from cold to hot data in order to achieve improved performance, and how numeric methods may be used with our model to find the optimum operating point, which approaches a write amplification of 1.0 for increasingly skewed traffic. We examine online methods for achieving this optimal operating point, and show that a control strategy based on our model achieves near-optimal performance for a number of real-world block traces.