SSD论文优秀句子

1. Nonvolatile memory(e.g., Phase Change Memory) blurs the boundary between memory and storage and it could greatly facilitate the construction of in-memory durable data structures. Data structures can be processed and stored directly in NVRAM. To XXX, YYY is a widely adopted mechanism. However, XXXXXXXX. By leveraging the XXXXX, we can YYYYYY. We tested our YYYYYY. Experiment results show that ZZZZZZZ, which can help extend the lifetime of NVRAM and improve performance.

2. By enabling efficient XXXX, YYY serve as the foundation for ZZZ.  For example: By enabling efficient insertions, point lookups, and range queries, key-value stores serve as the foundation for this growing group of important applications.

3. For write-intensive workloads, key-value stores based on Log-Structured Merge-Trees(LSM-trees)[1] have become the state of the art. Various distributed and local stores built on LSM-trees are widely deployed in large-scale production environments, such as BigTable [2] and LevelDB [3] at Google, Cassandra [4], HBase [5] and RocksDB [6] at Facebook, and Riak [7] at Basho. The main advantage of LSM-trees over other indexing structures (such as B-trees) is that they maintain sequential access patterns for writes. Small updates on B-trees may involve many random writes, and are hence not efficient on either solid-state storage devices or hard-disk drives.

-----

Reference

[1] Patrick ONeil, Edward Cheng, Dieter Gawlick, and Elizabeth ONeil. The Log-Structured MergeTree (LSM-tree). Acta Informatica, 33(4):351–385, 1996.

[2] Fay Chang, Jeffrey Dean, Sanjay Ghemawat, Wilson C. Hsieh, Deborah A. Wallach, Michael Burrows, Tushar Chandra, Andrew Fikes, and Robert Gruber. Bigtable: A Distributed Storage System for Structured Data. In Proceedings of the 7th Symposium on Operating Systems Design and Implementation (OSDI ’06), pages 205–218, Seattle, Washington, November 2006.

[3] Sanjay Ghemawat and Jeff Dean. LevelDB. http://code.google.com/p/leveldb, 2011.

[4] Avinash Lakshman and Prashant Malik. Cassandra – A Decentralized Structured Storage System. In The 3rd ACM SIGOPS International Workshop on Large Scale Distributed Systems and Middleware, Big Sky Resort, Montana, Oct 2009.

[5] Tyler Harter, Dhruba Borthakur, Siying Dong, Amitanand Aiyer, Liyin Tang, Andrea C. ArpaciDusseau, and Remzi H. Arpaci-Dusseau. Analysis of HDFS Under HBase: A Facebook Messages Case Study. In Proceedings of the 12th
USENIX Symposium on File and Storage Technologies (FAST ’14), Santa Clara, California, February 2014.

[6] Facebook. RocksDB. http://rocksdb.org/, 2013.

[7] Riak. http://docs.basho.com/riak/, 2015.

[8]