Improving Flash-Based Disk Cache with Lazy Adaptive Replacement

Huang, Sai; Wei, Qingsong; Feng, Dan; Chen, Jianxi; Chen, Cheng

doi:10.1145/2737832

Cited by 67 publications

(16 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, the trace search was collected from a popular search engine [25]. We compare our proposed ETD-Cache with two types of caching algorithms: 1) The classical cache schemes including LRU, MQ [15], LIRS [18], and ARC [19]; 2) The advanced schemes optimized for write reduction for SSD-based cache, including LARC [21] and L2ARC [22].…”

Section: Methodsmentioning

confidence: 99%

“…Some research work focused on this problem and put forward their settlement. For example, LARC [21] and SieveStore [20] only allow the blocks with close adjacent accesses or high enough access counts to enter SSD cache. Thus the quality of the cached data is promoted to maintain a high hit rate with less SSD writing requirements.…”

Section: Ssd Cost Index = Ssd Capacity × Average W Rite Bandwidthmentioning

confidence: 99%

“…The representative schemes include LARC [21] and SieveStore [20]. LARC adds a virtual LRU queue with limited length in front of the normal cache queue; the missed data blocks are first recorded in the newly introduced LRU queue.…”

Section: Cache Algorithmsmentioning

confidence: 99%

See 2 more Smart Citations

ETD-Cache

Dai

Chai

Liang

2015

Proceedings of the 12th ACM International Conference on Computing Frontiers

View full text Add to dashboard Cite

Recently flash-based solid-state drives (SSDs) have been widely deployed as cache devices to boost system performance. However, classical SSD cache algorithms (e.g. LRU) replace the cached data frequently to maintain high hit rates. Such aggressive data updating strategies result in too many writing operations on SSDs and make them wear out quickly, which finally leads to high costs of SSDs for enterprise applications. In this paper, we propose a novel Expiration-Time Driven Cache (ETD-Cache) method to solve this problem. In ETD-Cache, an active data eviction mechanism is adopted. An already cached block leaves the SSD cache if and only if there is no access to it for a time longer than a specified expiration time. This mechanism gives more time for the cached contents to wait for their following accesses and limits the admission of newly arrived blocks to generate less SSD writes. In addition, a low-overhead candidate management module is designed to maintain the most popular data in the system for the potential cache replacement. The simulations driven by a series of typical real-world traces indicate that due to the great reduction on data updating frequency, ETD-Cache lowers the total SSD costs by 98.45% compared with LRU under the same cache hit rate.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Ssd Cost Index = Ssd Capacity × Average W Rite Bandwidthmentioning

confidence: 99%

See 1 more Smart Citation

ETD-Cache

Dai

Chai

Liang

2015

Proceedings of the 12th ACM International Conference on Computing Frontiers

View full text Add to dashboard Cite

show abstract

“…RACS'15, October 9-12, 2015 Flashcache cache is a Linux cache device driver module developed by Facebook on a server with the HDD and SSD to use the SSD as a cache. It represents two block devices with speed difference (ex SSD, HDD) as a virtual device via Mapper [9][10]. All inputs and outputs for the device take place via cache device as RR (Round-Robin) policy.…”

Section: Releated Work 21 Hdfs Read I/omentioning

confidence: 99%

Design and implement of pre-loading SSD cache data using split file on Hadoop MapReduce

Kim

Jeon

2015

Proceedings of the 2015 Conference on Research in Adaptive and Convergent Systems

View full text Add to dashboard Cite

A storage device that is used mainly in the Hadoop system, HDD, has the slow transfer speed and long latency because of its physical limit. In addition, even though the SSD is the only alternative to traditional HDD, the price of it is too expensive. In order to improve this problem, some distributed file systems use SDD as a cache of HDD. In this paper, in the distributed file system that uses SSD as cache, we propose a mechanism that acquires a block ID that is scheduled to be used or already in use and loads it in SSD cache before the I/O request about the block is made in order to increase the performance of SSD cache during performance of Hadoop MapReduce task. In order to verify the reduction of MapReduce processing time and the improvement of Read I/O performance, which were run based on the proposed mechanism, experiments were carried out via Hadoop Benchmark tool. We could verify through the experiments that the performance of Hadoop MapReduce system was about 11% improved in the system where the proposed mechanism was applied than in the distributed file system where the existing SSD cache was applied. CCS Concepts• Computer systems organization➝Distributed architectures;• Software and its engineering➝Software performance.

show abstract

“…FlashTier [28] studied a new flash device interface specialized for caching, which reduces the block management overhead by unifying the block address mappings done by the cache and device and reduces the device garbage collection overhead by silently evicting clean cache blocks. HEC [30] and LARC [21] studied new cache admission policies to address the flash wear-out issue by not caching data that are infrequently used or from backup workloads. These solutions are complementary to this paper's study which focuses on the design issues internal to cache while our discoveries also have an impact on flash performance and endurance.…”

Section: Background and Motivationsmentioning

confidence: 99%

Client-side Flash Caching for Cloud Systems

Arteaga

Zhao

2014

Proceedings of International Conference on Systems and Storage

View full text Add to dashboard Cite

As the size of cloud systems and the number of hosted VMs rapidly grow, the scalability of shared VM storage systems becomes a serious issue. Client-side flash-based caching has the potential to improve the performance of cloud VM storage by employing flash storage available on the client-side of the storage system to exploit the locality inherent in VM IOs. However, because of the limited capacity and durability of flash storage, it is important to determine the proper size and configuration of the flash caches used in cloud systems. This paper provides answers to the key design questions of cloud flash caching based on dm-cache, a block-level caching solution customized for cloud environments, and a large amount of long-term traces collected from real-world public and private clouds. The study first validates that cloud workloads have good cacheability and dm-cache-based flash caching incurs low overhead with respect to commodity flash devices. It further reveals that write-back caching substantially outperforms write-through caching in typical cloud environments due to the reduction of server IO load. It also shows that there is a tradeoff on making a flash cache persistent across client restarts which saves hours of cache warm-up time but incurs considerable overhead from committing every metadata update persistently. Finally, to reduce the data loss risk from using write-back caching, the paper proposes a new cache-optimized RAID technique, which minimizes the RAID overhead by introducing redundancy of cache dirty data only, and shows to be significantly faster than traditional RAID and write-through caching.

show abstract

Improving Flash-Based Disk Cache with Lazy Adaptive Replacement

Cited by 67 publications

References 17 publications

ETD-Cache

ETD-Cache

Design and implement of pre-loading SSD cache data using split file on Hadoop MapReduce

Client-side Flash Caching for Cloud Systems

Contact Info

Product

Resources

About