Providing SLO Compliance on NVMe SSDs Through Parallelism Reservation
作者： Sheng-Min Huang , Li-Pin Chang
DOI: 10.1145/3174867
关键词:
摘要: Non-Volatile Memory Express (NVMe) is a specification for next-generation solid-state disks (SSDs). Benefited from the massive internal parallelism and high-speed PCIe bus, NVMe SSDs achieve extremely high data transfer rates, they are an ideal solution of shared storage in virtualization environments. Providing virtual machines with Service Level Objective (SLO) compliance on challenging task, because garbage collection activities inside globally affect I/O performance all machines. In this study, we introduce novel approach, called reservation, which inspired by rich SSDs. The degree stands how many flash chips concurrently active. Our basic idea to reserve sufficient degrees read, write, operations, making sure that SSD delivers stable read write throughput reclaims free space at constant rate. proportionally distributed among SLO compliance. experimental results show our reservation approach delivered satisfiable highly predictable response
acm.org
nctu.edu.tw参考文章(35)
Peter J. Varman, Ajay Gulati, Arif Merchant, mClock: handling throughput variability for hypervisor IO scheduling operating systems design and implementation. pp. 437- 450 ,(2010) , 10.5555/1924943.1924974
Mustafa Shihab, Myoungsoo Jung, Jie Zhang, Power, energy and thermal considerations in SSD-based I/O acceleration usenix conference on hot topics in storage and file systems. pp. 15- 15 ,(2014)
Sandeep Uttamchandani, Xiaosong Ma, Li Zhou, Fei Meng, Deng Liu, vCacheShare: automated server flash cache space management in a virtualization environment usenix annual technical conference. pp. 133- 144 ,(2014)
Jaeho Kim, Donghee Lee, Sam H. Noh, Towards SLO complying SSDs through OPS isolation file and storage technologies. pp. 183- 189 ,(2015) , 10.5555/2750482.2750496
Tae Yong Kim, Dong Hyun Kang, Dongwoo Lee, Young Ik Eom, Improving performance by bridging the semantic gap between multi-queue SSD and I/O virtualization framework ieee conference on mass storage systems and technologies. pp. 1- 11 ,(2015) , 10.1109/MSST.2015.7208295
Qi Zhang, Xuandong Li, Linzhang Wang, Tian Zhang, Yi Wang, Zili Shao, Optimizing deterministic garbage collection in NAND flash storage systems real time technology and applications symposium. pp. 14- 23 ,(2015) , 10.1109/RTAS.2015.7108392
Carl A. Waldspurger, Ajay Gulati, Irfan Ahmad, PARDA: proportional allocation of resources for distributed storage access file and storage technologies. pp. 85- 98 ,(2009)
Byunghei Jun, Dongkun Shin, Workload-aware budget compensation scheduling for NVMe solid state drives 2015 IEEE Non-Volatile Memory System and Applications Symposium (NVMSA). pp. 1- 6 ,(2015) , 10.1109/NVMSA.2015.7304369
Qiumin Xu, Huzefa Siyamwala, Mrinmoy Ghosh, Tameesh Suri, Manu Awasthi, Zvika Guz, Anahita Shayesteh, Vijay Balakrishnan, Performance analysis of NVMe SSDs and their implication on real world databases acm international conference on systems and storage. pp. 6- ,(2015) , 10.1145/2757667.2757684
Myoungsoo Jung, Mahmut T. Kandemir, Sprinkler: Maximizing resource utilization in many-chip solid state disks high-performance computer architecture. pp. 524- 535 ,(2014) , 10.1109/HPCA.2014.6835961