Search

For a production high-performance computing (HPC) system, where storage devices are shared between multiple applications and managed in a best effort manner, I/O contention is often a major problem. In this paper, we propose a balanced messaging-based re-routing in conjunction with throttling at the middleware level. This work tackles two key challenges that have not been fully resolved in the past: whether I/O variability can be reduced on a QoS-less HPC storage system, and how to design a runtime scheduling system that can scale up to a large amount of cores. The proposed scheme uses a two-level messaging system to re-route I/O requests to a less congested storage location so that write performance is improved, while limiting the impact on read by throttling re-routing. An analytical model is derived to guide the setup of optimal throttling factor. We thoroughly analyze the virtual messaging layer overhead and explore whether the in-transit buffering is effective in managing I/O variability. Contrary to the intuition, in-transit buffer cannot completely solve the problem. It can reduce the absolute variability but not the relative variability. The proposed scheme is verified against a synthetic benchmark as well as being used by production applications. [ABSTRACT FROM AUTHOR]

DRAM memory suffers from increasingly aggravating refresh penalty, which causes significant performance degradation and power consumption. As memory capacity increases, refresh penalty has become increasingly worse as more rows have to be refreshed. In this work, we propose an effective refresh approach called Compression-Aware Refresh (CAR) to efficiently mitigate refresh overheads. We apply a data compression technique to store data in a compressed format so that the resultant sparse banks need only be partially refreshed. Because of compression, data blocks which are originally distributed across all the constituent chips of a rank only need to be stored in a subset of those chips, leaving banks in the remaining chips not fully occupied. As a result, the memory controller can safely skip refreshing memory rows which contain no useful data without compromising data integrity. Such a compression-aware refresh scheme significantly reduces the refresh and thus improves overall memory performance and energy efficiency. To further take advantage of data compression, we adopt a rank subsetting technique to enable accesses to only those occupied chips for memory requests accessing compressed data blocks. Evaluations using benchmarks from SPEC CPU 2006 and the PARSEC 3.0 on recent DDR4 memory systems have shown that CAR achieves up to 1.66x performance improvement (11.7 percent on average) and up to 45.9 percent energy reduction (27.3 percent on average), and reduce memory traffic by up to 99.9 percent for zero cache lines intensive workloads with an average of 66.3 percent across all benchmarks. [ABSTRACT FROM AUTHOR]

In backup systems, the chunks of each backup are physically scattered after deduplication, which causes a challenging fragmentation problem. We observe that the fragmentation comes into sparse and out-of-order containers. The sparse container decreases restore performance and garbage collection efficiency, while the out-of-order container decreases restore performance if the restore cache is small. In order to reduce the fragmentation, we propose History-Aware Rewriting algorithm (HAR) and Cache-Aware Filter (CAF). HAR exploits historical information in backup systems to accurately identify and reduce sparse containers, and CAF exploits restore cache knowledge to identify the out-of-order containers that hurt restore performance. CAF efficiently complements HAR in datasets where out-of-order containers are dominant. To reduce the metadata overhead of the garbage collection, we further propose a Container-Marker Algorithm (CMA) to identify valid containers instead of valid chunks. Our extensive experimental results from real-world datasets show HAR significantly improves the restore performance by 2.84-175.36 $\times$ at a cost of only rewriting 0.5-2.03 percent data. [ABSTRACT FROM PUBLISHER]

Data deduplication has become a commodity in large-scale storage systems, especially in data backup and archival systems. However, due to the removal of redundant data, data deduplication de-linearizes data placement and forces the data chunks of the same data object to be divided into multiple separate units. In our preliminary study, we found that the de-linearization of data placement compromises the data spatial locality that is used to improve data read performance, deduplication throughput and deduplication efficiency in some deduplication approaches, which significantly affects deduplication performance and makes some deduplication approaches become less effective. In this paper, we first analyze the negative effect of data placement de-linearization to deduplication performance, and then propose an effective approach called De-Frag to reduce the de-linearization of data placement. The key idea of De-Frag is to choose some redundant data to be written to the disks rather than be removed. It quantifies the spatial locality of each chunk group by spatial locality level (SPL for short) and writes the redundant chunks to disks when SPL value is smaller than a preset value, thus to reduce the de-linearization of data placement and enhance the spatial locality. As shown in our experimental results driven by real world datasets, De-Frag effectively enhances data spatial locality and improves deduplication throughput, deduplication efficiency, and data read performance, at the cost of slightly lower compression ratios. [ABSTRACT FROM AUTHOR]

I/O performance has been critical for large-scale distributed systems. Many approaches, including hint-based multilevel cache, have been proposed to smooth the gap between different levels. These solutions demote or promote cache blocks based on the latest history information, which is insufficient for applications where frequent demote and promote operations occur. In this paper, we propose a novel multilevel buffer cache using K-step hints (Hint-K) to improve the I/O performance of distributed systems. The basic idea is to promote a block from the lower level cache to the higher level(s) or demote a block vice versa based on the block's previous K-step promote or demote operations, which are referred to as K-step hints. If we make an analogy between Hint-K and LRU-K, then LRU-K keeps track of the times of last K references for blocks within a single cache level, while our Hint-K keeps track of the information of the last K movements (either demote or promote) of blocks among different cache levels. We develop our Hint-K algorithms and design a mathematical model that can efficiently describe the activeness of any block in any cache level. Simulation results show that Hint-K achieves better performance compared to the existing popular multilevel cache schemes such as PROMOTE, DEMOTE, and MQ under different I/O workloads. [ABSTRACT FROM AUTHOR]

Hybrid co-scheduling (CS) is an effective approach to address the synchronisation problems for concurrent virtual machines such as lock-holder preemption. However, the scheduling gaps and fragments in hybrid CS cause a serious performance degradation and unfairness if multiple concurrent domains exist. We propose three optimisation schemes for such problems named partial co-scheduling (PCS), boost co-scheduling (BCS) and precise boost co-scheduling (PBCS) using finer space granularity. PCS scheme activates the CS signals only for the indispensable central processing units (CPUs) instead of all online CPUs. BCS scheme boosts the priorities for co-scheduled virtual CPUs (VCPUs) to induce the scheduler to pick the appropriate VCPUs. PBCS combines PCS and BCS to achieve better performance and robustness. We implement three optimisations into Credit Scheduler in Xen 4.0.1. The experimental results show that our schemes achieve better performance and fairness compared to existing hybrid CS under different scenarios. [ABSTRACT FROM PUBLISHER]

Abstract: In NAND flash memory, once a page program or block erase (P/E) command is issued to a NAND flash chip, the subsequent read requests have to wait until the time-consuming P/E operation to complete. Preliminary results show that the lengthy P/E operations increase the read latency by 2× on average. This increased read latency caused by the contention may significantly degrade the overall system performance. Inspired by the internal mechanism of NAND flash P/E algorithms, we propose in this paper a low-overhead P/E suspension scheme, which suspends the on-going P/E to service pending reads and resumes the suspended P/E afterwards. Having reads enjoy the highest priority, we further extend our approach by making writes be able to preempt the erase operations in order to improve the write latency performance. In our experiments, we simulate a realistic SSD model that adopts multi-chip/channel and evaluate both SLC and MLC NAND flash as storage materials of diverse performance. Experimental results show the proposed technique achieves a near-optimal performance on servicing read requests. The write latency is significantly reduced as well. Specifically, the read latency is reduced on average by 46.5% compared to RPS (Read Priority Scheduling) and when using write–suspend–erase the write latency is reduced by 13.6% relative to FIFO. [Copyright &y& Elsevier]