• International Journal of Internet, Broadcasting and Communication
• /
• v.13 no.2
• /
• pp.27-35
• /
• 2021

NAND flash memory-based SSD needs an internal software, Flash Translation Layer(FTL) to provide traditional block device interface to the host because of its physical constraints, such as erase-before-write and large erase block. However, because useful host-side information cannot be delivered to FTL through the narrow block device interface, SSDs suffer from a variety of problems such as increasing garbage collection overhead, large tail-latency, and unpredictable I/O latency. Otherwise, the new type of SSD, open-channel SSD exposes the internal structure of SSD to the host so that underlying NAND flash memory can be managed directly by the host-level FTL. Especially, I/O data classification by using host-side information can achieve the reduction of garbage collection overhead. In this paper, we propose a new scheme to reduce garbage collection overhead of open-channel SSD by separating the journal from other file data for the journaling filesystem. Because journal has different lifespan with other file data, the Write Amplification Factor (WAF) caused by garbage collection can be reduced. The proposed scheme is implemented by modifying the host-level FTL of Linux and evaluated with both Fio and Filebench. According to the experiment results, the proposed scheme improves I/O performance by 46%~50% while reducing the WAF of open-channel SSDs by more than 33% compared to the previous one.

• The Journal of the Korea Contents Association
• /
• v.18 no.4
• /
• pp.178-185
• /
• 2018

Recently, demands for computer systems with multicore CPUs and high-performance flash-based storage devices (i.e., flash SSD) have rapidly grown in cloud computing, surer-computing, and enterprise storage/database systems. Journaling file systems running on high-performance systems do not exploit the full I/O bandwidth of high-performance SSDs. In this article, we evaluate and analyze the performance of the Linux EXT4 file system with high-performance SSDs and multicore CPUs. The system used in this study has 72 cores and Intel NVMe SSD, and the flash SSD has performance up to 2800/1900 MB/s for sequential read/write operations. Our experimental results show that checkpointing in the EXT4 file system is a major overhead. Furthermore, we optimize the checkpointing procedure and our optimized EXT4 file system shows up to 92% better performance than the original EXT4 file system.

• ETRI Journal
• /
• v.39 no.6
• /
• pp.820-831
• /
• 2017

File systems and applications try to implement their own update protocols to guarantee data consistency, which is one of the most crucial aspects of computing systems. However, we found that the storage devices are substantially under-utilized when preserving data consistency because they generate massive storage write traffic with many disk cache flush operations and force-unit-access (FUA) commands. In this paper, we present DJFS (Delta-Journaling File System) that provides both a high level of performance and data consistency for different applications. We made three technical contributions to achieve our goal. First, to remove all storage accesses with disk cache flush operations and FUA commands, DJFS uses small-sized NVRAM for a file system journal. Second, to reduce the access latency and space requirements of NVRAM, DJFS attempts to journal compress the differences in the modified blocks. Finally, to relieve explicit checkpointing overhead, DJFS aggressively reflects the checkpoint transactions to file system area in the unit of the specified region. Our evaluation on TPC-C SQLite benchmark shows that, using our novel optimization schemes, DJFS outperforms Ext4 by up to 64.2 times with only 128 MB of NVRAM.

• Journal of IKEEE
• /
• v.23 no.1
• /
• pp.80-90
• /
• 2019

Even though Non-Volatile Memory (NVM) is used for data storage, a page table should be built to access data in it. And this observation leads us to the Persistent Page Table (PPT) scheme that keeps the page table in NVM persistently. By the way, processors have different page table structures and really operational page table cannot be built without virtual and physical addresses of NVM. However, those addresses are determined dynamically when NVM storage is attached to the system. Thus, the PPT should have system-independent and also address-independent structure and really working system-dependent page table should be built from the PPT. Moreover, entries of PPT should be updated atomically and, in this paper, we describe the design of PPT that meets those requirements. And we investigate how file systems can decrease the journaling overhead with the swap operation, which is a new operation created by the PPT. We modified the Ext4 file system in Linux and experiments conducted with Filebench workloads show that the swap operation enhances file system performance up to 60%.

• International journal of advanced smart convergence
• /
• v.10 no.1
• /
• pp.142-150
• /
• 2021

Open-channel SSD is a new type of Solid-State Disk (SSD) that improves the garbage collection overhead and write amplification due to physical constraints of NAND flash memory by exposing the internal structure of the SSD to the host. However, the host-level Flash Translation Layer (FTL) provided for open-channel SSDs in the current Linux kernel consumes host memory excessively because it use page-level mapping table to translate logical address to physical address. Therefore, in this paper, we implemente a selective mapping table loading scheme that loads only a currently required part of the mapping table to the mapping table cache from SSD instead of entire mapping table. In addition, to increase the hit ratio of the mapping table cache, filesystem information and mapping table access history are utilized for cache replacement policy. The proposed scheme is implemented in the host-level FTL of the Linux kernel and evaluated using open-channel SSD emulator. According to the evaluation results, we can achieve 80% of I/O performance using the only 32% of memory usage compared to the previous host-level FTL.

• The Journal of Korean Association of Computer Education
• /
• v.17 no.2
• /
• pp.97-106
• /
• 2014

There is a lot of interest in the data storage and backup systems according to increasing the data intensive services and related user's data. The overhead of backup performance in massive storage system is a critical issue because the traditional incremental backup strategies causes the time consuming bottleneck in the SAN environment. The SANique Smart Vault system is a high performance backup solution with data de-duplication technology and it guarantees these requirements. In this paper, we describe the architecture of SANique Smart Vault system and illustrate efficient delta incremental backup method based on journaling files. We also present the record-level data de-duplication method in our proposed backup system. The proposed forever incremental backup and data de-duplication algorithms are analyzed and investigated by performance evaluation of other commercial backup solutions.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.41 no.2
• /
• pp.153-158
• /
• 2018

At the Bank of Korea, capital stock statistics were created by the PIM (perpetual inventory method) with fixed capital formation data. Asset classifications also included 2 categories in residential buildings, 4 non-residential buildings, 14 constructions, 9 transportation equipment, 28 machinery, and 2 intangible fixed assets. It is the Korean government accounting system which is developed much with the field of the national accounts including the valuation, but until 2008 it was consistent with single-entry bookkeeping. Many countries, including Korea, were single-entry bookkeeping, not double-entry bookkeeping which can be aggregated by government accounting standard account. There was no distinction in journaling between revenue and capital expenditure when it was consistent with single-entry bookkeeping. For example, we would like to appropriately divide the past budget accounts and the settlement accounts data that have been spent on dredging into capital expenditure and revenue expenditure. It, then, tries to add the capital expenditure calculated to FCF (fixed capital formation), because revenue expenditure is cost for maintenance etc. This could be a new direction, especially, in the estimation of capital stock by the perpetual inventory method for infrastructure (SOC, social overhead capital). It should also be noted that there are differences not only between capital and income expenditure but also by other factors. How long will this difference be covered by the difference between the 'new series' and 'old series' methodologies? In addition, there is no large difference between two series by the major asset classification level. If this is treated as a round-off error, this is a problem.