Improving flash write performance by using update frequency

Abstract

Solid-state drives (SSDs) are quickly becoming the default storage medium as the cost of NAND flash memory continues to drop. However, flash memory introduces new challenges, as data cannot be efficiently updated in-place. To overcome the technology's limitations, SSDs incorporate a software Flash Translation Layer (FTL) that implements out-of-place updates, typically by storing data in a log-structured fashion. Despite a large number of existing FTL algorithms, SSD performance, predictability, and lifetime remain an issue, especially for the write-intensive workloads specific to database applications.In this paper, we show how to design FTLs that are more efficient by using the I/O write skew to guide data placement on flash memory. We model the relationship between data placement and write performance for basic I/O write patterns and detail the most important concepts of writing to flash memory: i) the trade-off between the extra capacity available and write overhead, ii) the benefit of adapting data placement to write skew, iii) the impact of the cleaning policy, and iv) how to estimate the best achievable write performance for a given I/O workload. Based on the findings of the theoretical model, we propose a new principled data placement algorithm that can be incorporated into existing FTLs. We show the benefits of our data placement algorithm when running micro-benchmarks and real database I/O traces: our data placement algorithm reduces write overhead by 20% -75% when compared to state-of-art techniques.

A Preliminary Study: Towards Parallel Garbage Collection for NAND Flash-Based SSDs

“…It separates cold and hot valid pages during GC, which reduces the page copy overhead. Our study is in line with these works [33], [46], [52], [54], [56] in terms of improving SSD performance. In contrast, we focus on improving the internal GC procedure.…”

“…Hot/cold data separation: Lee et al [33] perform an empirical study that shows the hot/cold data separation policies which can reduce the GC overhead. Stoica et al [46] proposed an update frequency-based data placement algorithm. This algorithm reduces the number of valid pages within a victim block, which reduces the GC overhead.…”

Online Updates on Data Warehouses via Judicious Use of Solid-State Storage

“…Note that the devices used have higher SSD capacity-that is, 32GB and 64GB-but the MaSM algorithms need only a fraction of that space to offer high update rate with low or negligible impact on the query response time. Hence the rest of the SSD capacity can be used by the database system for other operations, or can be used as additional space, allowing the SSD device to optimize garbage collection and prolong its lifetime [Bux and Iliadis 2010;Stoica and Ailamaki 2013]. Similarly, the MaSM algorithms need a fraction of the available memory for its operation, leaving the better part of the memory available for memory-hungry operators like hash-joins.…”

Incremental redundancy to reduce data retention errors in flash-based SSDs

“…In particular, we assume that data are updated randomly with the same probability and that the LRW (Least Recently Written) policy is used to select victim blocks for GC [25]. For further derivations of GC period and effective OPS (Over Provisioning Space) size in the next section, we briefly present results of a previous study that derives the utilization of the victim block for GC [37].…”

“…Assuming that U is extremely large, u can be transformed by applying the Euler's limit and the Lambert-W function as follows [37], [38]:…”