Storage Hierarchy Research Articles

A New Memory-Disk Integrated System with HW Optimizer

Current high-performance computer systems utilize a memory hierarchy of on-chip cache, main memory, and secondary storage due to differences in device characteristics. Limiting the amount of main memory causes page swap operations and duplicates data between the main memory and the storage device. The characteristics of next-generation memory, such as nonvolatility, byte addressability, and scaling to greater capacity, can be used to solve these problems. Simple replacement of secondary storage with new forms of nonvolatile memory in a traditional memory hierarchy still causes typical problems, such as memory bottleneck, page swaps, and write overhead. Thus, we suggest a single architecture that merges the main memory and secondary storage into a system called a Memory-Disk Integrated System (MDIS). The MDIS architecture is composed of a virtually decoupled NVRAM and a nonvolatile memory performance optimizer combining hardware and software to support this system. The virtually decoupled NVRAM module can support conventional main memory and disk storage operations logically without data duplication and can reduce write operations to the NVRAM. To increase the lifetime and optimize the performance of this NVRAM, another hardware module called a Nonvolatile Performance Optimizer (NVPO) is used that is composed of four small buffers. The NVPO exploits spatial and temporal characteristics of static/dynamic data based on program execution characteristics. Enhanced virtual memory management and address translation modules in the operating system can support these hardware components to achieve a seamless memory-storage environment. Our experimental results show that the proposed architecture can improve execution time by about 89% over a conventional DRAM main memory/HDD storage system, and 77% over a state-of-the-art PRAM main memory/HDD disk system with DRAM buffer. Also, the lifetime of the virtually decoupled NVRAM is estimated to be 40% longer than that of a traditional hierarchy based on the same device technology.

Tier 3 batch system data locality via managed caches

Modern data processing increasingly relies on data locality for performance and scalability, whereas the common HEP approaches aim for uniform resource pools with minimal locality, recently even across site boundaries. To combine advantages of both, the High- Performance Data Analysis (HPDA) Tier 3 concept opportunistically establishes data locality via coordinated caches.In accordance with HEP Tier 3 activities, the design incorporates two major assumptions: First, only a fraction of data is accessed regularly and thus the deciding factor for overall throughput. Second, data access may fallback to non-local, making permanent local data availability an inefficient resource usage strategy. Based on this, the HPDA design generically extends available storage hierarchies into the batch system. Using the batch system itself for scheduling file locality, an array of independent caches on the worker nodes is dynamically populated with high-profile data. Cache state information is exposed to the batch system both for managing caches and scheduling jobs. As a result, users directly work with a regular, adequately sized storage system. However, their automated batch processes are presented with local replications of data whenever possible.

Electrical determination of vortex state in submicron magnetic elements

Vortex structures in confined geometries are currently under close scrutiny due to their unique properties associated with their spatial confinement and the non-uniform distribution of the magnetization. The magnetic vortex is characterized by two boolean topological quantities: circulation (clockwise or counterclockwise c = ± 1) of the in-plane magnetization and polarity (up or down, p = ± 1) of the vortex core. These four degenerate states are quite stable and can accelerate the development of more compact and high performance magnetic memory devices. Thus an understanding of their dynamical behavior and a way to electrically detect these states is a major requirement for their development. Progress can be achieved by combining theoretical calculations, micromagnetic simulations and experimental approaches. The phenomenon of spin transfer torque is exploited to excite the lowest frequency (gyro) mode of the vortex core confined in a submicron magnetic (Permalloy - Ni81Fe19) element. The gyrotropic motion of the vortex core leads to a periodic change in the magnetization and hence its resistance: due to the anisotropic magnetoresistance (AMR) effect. This periodic change in resistance combines with the excitation current and generates a periodic homodyne voltage signal. An external static magnetic field is applied to break the symmetry and to rectify the homodyne voltage signal which we measure in a nanovoltmeter. It is found that the sign of the rectified AMR signal depends upon the handedness (cp) of the vortex structure. Micromagnetic simulations provide better understanding and are in good agreement with our experimental results. Additionally, vortex dynamics in these samples is investigated in a Scanning Transmission X-ray Microscope (STXM) with a temporal (< 100 ps) and spatial (~ 30 nm) resolution which allows us to verify the resonance frequency of the magnetic element as well as the power range to excite the vortex core. The AMR based technique thus can be used to detect the circulation and the polarity of the vortex state electrically and could open a route to implement magnetic vortex elements in memory and storage hierarchies. The phenomenon of Spin Motive Force (SMF) has also been studied by micromagnetic simulations. It is found that, in a particular configuration, the SMF signal shows a phase difference of 180 degrees for two polarities of the vortex core, when the voltage probe contacts are located parallel to the excitation rf field direction. No phase shift is observed in the perpendicular case. In addition, a 180 degree phase difference is observed for different circulations of the vortex structure. Therefore, this could also be a possible way to determine polarity and circulation of the magnetic vortex by carefully examining the phase relation of the SMF generated voltage signals. An attempt has also been made to measure the SMF experimentally. However, due to the small expected signal unambiguous detection of SMF was not successful so far.

IOPro: a parallel I/O profiling and visualization framework for high-performance storage systems

Efficient execution of large-scale scientific applications requires high-performance computing systems designed to meet the I/O requirements. To achieve high-performance, such data-intensive parallel applications use a multi-layer layer I/O software stack, which consists of high-level I/O libraries such as PnetCDF and HDF5, the MPI library, and parallel file systems. To design efficient parallel scientific applications, understanding the complicated flow of I/O operations and the involved interactions among the libraries is quintessential. Such comprehension helps identify I/O bottlenecks and thus exploits the potential performance in different layers of the storage hierarchy. To profile the performance of individual components in the I/O stack and to understand complex interactions among them, we have implemented a GUI-based integrated profiling and analysis framework, IOPro. IOPro automatically generates an instrumented I/O stack, runs applications on it, and visualizes detailed statistics based on the user-specified metrics of interest. We present experimental results from two different real-life applications and show how our framework can be used in practice. By generating an end-to-end trace of the whole I/O stack and pinpointing I/O interference, IOPro aids in understanding I/O behavior and improving the I/O performance significantly.

Design of Secure Cloud Storage System for Enterprise Information Resources

In order to meet growing demand of storage resources, improve the security and integrity of enterprise information resources, this paper analyzes storage resources status and existing problems of Foshan XY Elec-Mech Limited Corp in the first. Based on detailed analysis of cloud storage hierarchy and its advantages, the cloud storage framework is designed to meet enterprise’s demands, security control strategies and backup strategies are also discussed. After implementation, the integration of information resources in multi-systems and multi-platforms may realized for providing users with massive information storage and access services. Furthermore, when disaster coming, multiple backups may be used to recovery for ensuring the continuity of enterprise’s business processing.

Applied Research for Campus Student Credit Management System under the Cloud Storage

Campus student credit management system under cloud storage is studied to improve performance of management system. In universities, credit is an important indicator of student achievement, and a direct impact on whether students can graduate. How to manage credit of students, has become a key topic of campus management. To this end, the campus student credit management systemunder cloud storage environmentsis proposed. The relevant principles and hierarchy of cloud storage are elaborated, and storage environment of campus student credit management system is obtained. In these environments, student credit management is achieved. Experimental results show that with the improved algorithm to process campus student credit management under cloud storage, the storage space can be expanded to improve management efficiency, to meet the actual needs of the campus administration.

Adapting Server Systems for New Memory Technologies

After decades of a cache, DRAM, and disk data storage hierarchy, new memory technologies are promising nonvolatility, high endurance, and fine-grained data access--all of which could, with some preparation, enable a new storage paradigm in server systems.

Design and Prototype of a Solid-State Cache

The availability of high-speed solid-state storage has introduced a new tier into the storage hierarchy. Low-latency and high-IOPS solid-state drives (SSDs) cache data in front of high-capacity disks. However, most existing SSDs are designed to be a drop-in disk replacement, and hence are mismatched for use as a cache. This article describes FlashTier , a system architecture built upon a solid-state cache (SSC), which is a flash device with an interface designed for caching. Management software at the operating system block layer directs caching. The FlashTier design addresses three limitations of using traditional SSDs for caching. First, FlashTier provides a unified logical address space to reduce the cost of cache block management within both the OS and the SSD. Second, FlashTier provides a new SSC block interface to enable a warm cache with consistent data after a crash. Finally, FlashTier leverages cache behavior to silently evict data blocks during garbage collection to improve performance of the SSC. We first implement an SSC simulator and a cache manager in Linux to perform an in-depth evaluation and analysis of FlashTier's design techniques. Next, we develop a prototype of SSC on the OpenSSD Jasmine hardware platform to investigate the benefits and practicality of FlashTier design. Our prototyping experiences provide insights applicable to managing modern flash hardware, implementing other SSD prototypes and new OS storage stack interface extensions. Overall, we find that FlashTier improves cache performance by up to 168% over consumer-grade SSDs and up to 52% over high-end SSDs. It also improves flash lifetime for write-intensive workloads by up to 60% compared to SSD caches with a traditional flash interface.

Profit data caching and hybrid disk-aware Completely Fair Queuing scheduling algorithms for hybrid disks

Recently, a hybrid disk drive that integrates a small amount of flash memory within a mechanical drive has received significant attention. The hybrid drive extends the storage hierarchy by using flash memory to cache data from the mechanical disk. Unfortunately, current caching architectures fail to fully exploit the potential of the hybrid drive. Furthermore, current disk input/output I/O schedulers are optimized for rotational mechanical disk drives and thus must be re-targeted for the hybrid disk drive. In this paper, we propose a new data caching scheme, called Profit Caching, for hybrid drives. Profit Caching is a self-optimizing caching algorithm. It considers and seamlessly integrates all possible data characteristics that impact the performance of hybrid drives, including read count, write count, sequentiality, randomness, and recency, to determine the caching policy. Moreover, we propose a hybrid disk-aware Completely Fair Queuing HA-CFQ scheduler to avoid unnecessary I/O anticipations of the CFQ scheduler. We have implemented Profit Caching and HA-CFQ scheduler in the Linux kernel. Coupled with a trace-driven simulator, we have also conducted detailed experiments under a variety of workloads. Experimental results show that Profit Caching provides significantly improved performance compared with the previous schemes. In particular, the throughput of Profit Caching outperforms previous Random Access First and FlashCache caching schemes by factors of up to 1.8 and 7.6, respectively. In addition, the HA-CFQ scheduler reduces the total execution time of the CFQ scheduler by up to 1.74%. Finally, the experimental results show that the runtime overhead of Profit Caching is extremely insignificant and can be ignored. Copyright © 2014 John Wiley & Sons, Ltd.

Towards Adapting NAS Mechanism over Solid State Disks

Flash memory storage such as SSDs (Solid State disks) have been gaining popularity due to its low energy consumption and durability in embedded systems and laptops. With the fast technical improvement, Solid State Disks (SSDs) are becoming an important part of the computer storage hierarchy to significantly improve performance and energy efficiency. However, due to its relatively high price and low capacity, a major system research issue to address is on how to make SSDs play their most effective roles as a high-performance storage system in cost- and performance-effective ways. In this paper, we describe a NAS prototype that eliminates most software overheads and makes Solid State Disks a high-performance storage system in cost- and performance-effective ways. NAS integrates a network adapter into SSDs, so the SSDs remove the random access time, Reduce administrative overhead while delivering scalable, reliable data storage. NAS can also give support block based I/O rather than traditional NAS system which supports only file I/O protocol such as NFS and CIFS. Our scale-out NAS solutions can boost capacity and performance while enabling flexible provisioning, no disruptive operations, easy scalability, and virtualization. As the storage and processing requirements of the file server continued to increase data security and integrity became difficult to

A lightweight software fault‐tolerance system in the cloud environment

SummaryWith the development of cloud computing, the demand of high availability for services is growing. Unfortunately, software failures greatly reduce system availability. This paper presents a lightweight software fault‐tolerance system, called SHelp, which can effectively recover programs from many types of software bugs in the cloud environment. With error virtualization techniques, it proposes ‘weighted’ rescue points techniques to effectively survive software failures through bypassing the faulty path. For multiple application instances running on different virtual machine, a three‐level storage hierarchy with several comprehensive cache updating algorithms for rescue points management is adopted to share error handling information. On the one hand, SHelp can reduce the redundancy for multiple application instances; on the other hand, it can more effectively and quickly recover from faults caused by the same bugs. A Linux prototype is implemented on an open‐source virtual machine monitor platform, Xen, and evaluated using four Web server applications that contain various types of bugs. The experimental results show that SHelp can recover server applications from these bugs in just a few seconds with modest performance overhead. Copyright © 2013 John Wiley &amp; Sons, Ltd.

Frequency and diameter dependent viscoelastic properties of mitral valve chordae tendineae

This study aimed to characterise viscoelastic properties of different categories of chordae tendineae over a range of frequencies. Dynamic Mechanical Analysis (DMA) was performed using a materials testing machine. Chordae (n=51) were dissected from seven porcine hearts and categorised as basal, marginal, strut or commissural. Chordae were loaded under a sinusoidally varying tensile load at a range of frequencies between 0.5 and 5Hz, both at a standardised load (i.e. same mean load of 4N for all chordae) and under chordal specific loading (i.e. based on in vivo loads for different chordae). Storage modulus and stiffness were frequency-dependent. Loss modulus and stiffness were frequency-independent. Storage and loss moduli, but not stiffness, decreased with chordal diameter. Therefore, strut chordae have the lowest moduli and marginal chordae the highest moduli. The hierarchy of dynamic storage and loss moduli is: marginal, commissural, basal and strut. In conclusion, viscoelastic properties of chordae are dependent on both frequency and chordal type. Future/novel replacement chordal materials should account for frequency and diameter dependent viscoelastic properties of chordae tendineae.

Anti-caching

The traditional wisdom for building disk-based relational database management systems (DBMS) is to organize data in heavily-encoded blocks stored on disk, with a main memory block cache. In order to improve performance given high disk latency, these systems use a multi-threaded architecture with dynamic record-level locking that allows multiple transactions to access the database at the same time. Previous research has shown that this results in substantial overhead for on-line transaction processing (OLTP) applications [15]. The next generation DBMSs seek to overcome these limitations with architecture based on main memory resident data. To overcome the restriction that all data fit in main memory, we propose a new technique, called anti-caching, where cold data is moved to disk in a transactionally-safe manner as the database grows in size. Because data initially resides in memory, an anti-caching architecture reverses the traditional storage hierarchy of disk-based systems. Main memory is now the primary storage device. We implemented a prototype of our anti-caching proposal in a high-performance, main memory OLTP DBMS and performed a series of experiments across a range of database sizes, workload skews, and read/write mixes. We compared its performance with an open-source, disk-based DBMS optionally fronted by a distributed main memory cache. Our results show that for higher skewed workloads the anti-caching architecture has a performance advantage over either of the other architectures tested of up to 9× for a data size 8× larger than memory.

HybridPlan: a capacity planning technique for projecting storage requirements in hybrid storage systems

Economic forces, driven by the desire to introduce flash into the high-end storage market without changing existing software-base, have resulted in the emergence of solid-state drives (SSDs), flash packaged in HDD form factors and capable of working with device drivers and I/O buses designed for HDDs. Unlike the use of DRAM for caching or buffering, however, certain idiosyncrasies of NAND Flash-based solid-state drives (SSDs) make their integration into hard disk drive (HDD)-based storage systems nontrivial. Flash memory suffers from limits on its reliability, is an order of magnitude more expensive than the magnetic hard disk drives (HDDs), and can sometimes be as slow as the HDD (due to excessive garbage collection (GC) induced by high intensity of random writes). Given the complementary properties of HDDs and SSDs in terms of cost, performance, and lifetime, the current consensus among several storage experts is to view SSDs not as a replacement for HDD, but rather as a complementary device within the high-performance storage hierarchy. Thus, we design and evaluate such a hybrid storage system with HybridPlan that is an improved capacity planning technique to administrators with the overall goal of operating within cost-budgets. HybridPlan is able to find the most cost-effective hybrid storage configuration with different types of SSDs and HDDs

Making updates disk-I/O friendly using SSDs

Multiversion databases store both current and historical data. Rows are typically annotated with timestamps representing the period when the row is/was valid. We develop novel techniques for reducing index maintenance in multiversion databases, so that indexes can be used effectively for analytical queries over current data without being a heavy burden on transaction throughput. To achieve this end, we re-design persistent index data structures in the storage hierarchy to employ an extra level of indirection. The indirection level is stored on solid state disks that can support very fast random I/Os, so that traversing the extra level of indirection incurs a relatively small overhead. The extra level of indirection dramatically reduces the number of magnetic disk I/Os that are needed for index updates, and localizes maintenance to indexes on updated attributes. Further, we batch insertions within the indirection layer in order to reduce physical disk I/Os for indexing new records. By reducing the index maintenance overhead on transactions, we enable operational data stores to create more indexes to support queries. We have developed a prototype of our indirection proposal by extending the widely used Generalized Search Tree (GiST) open-source project, which is also employed in PostgreSQL. Our working implementation demonstrates that we can significantly reduce index maintenance and/or query processing cost, by a factor of 3. For insertions of new records, our novel batching technique can save up to 90% of the insertion time.

Operating systems and runtime environments on supercomputers

The evolving architecture of high-performance computing nodes poses new challenges to low-level system software components running on the nodes. New multicore and many-core designs put strain on system scalability; the introduction of nonvolatile memory forces a rethink of the memory and storage hierarchy in extreme-scale systems; the overall increase in the component count requires careful consideration of power management and fault tolerance approaches. The International Workshop on Runtime and Operating Systems for Supercomputers (ROSS), held in cooperation with the ACM Special Interest Group on High Performance Computing (SIGHPC), provides a forum for researchers to exchange ideas and discuss research questions that are relevant to low-level system software on current and future-generation supercomputers. This special issue presents significantly extended versions of the best papers from the 2013 edition of the workshop, which was held in conjunction with the ACM International Conference on Supercomputing (ICS) in Eugene, Oregon, USA, on June 10, 2013. In his keynote address Characteristics of Adaptive Runtime Systems in HPC, Sanjay Kale examined the features that are necessary or desirable to empower adaptive runtime systems, using Charm++ and the Adaptive Message Passing interface (MPI) as examples. Nine regular papers and one invited talk were also presented at the workshop in three sessions on exascale and energy, memory and tasking, and big data. Levy et al., in their paper A Study of the Viability of Exploiting Memory Content Similarity to Improve Resilience to Memory Errors, propose a novel runtime for transparently exploiting memory content similarity to reduce the rate of fatal memory errors by examining memory snapshots from HPC operating systems and applications. In Collective I/O under Memory Constraints, Lu et al. introduce a new collective I/O strategy to combat the problem of the decreasing amount of memory available per CPU core, by considering the capacity and bandwidth constraints, restricting aggregation data traffic, coordinating I/O accesses, and choosing aggregators at run time. These research efforts identify major challenges facing OS and runtime system developers and propose promising solutions to address them. We hope that the principles and techniques presented in this special issue will play an important role in the design and development of specialized operating systems for the exascale era.

Large-scale fast Fourier transform on a heterogeneous multi-core system

As interest in hybrid computing systems increases, people are eager to find new ways to exploit the unique and efficient computational power of the heterogeneous multi-core systems. Although there has been much interest in implementing high-performance fast Fourier transform (FFT) libraries on this kind of system, most existing libraries focus on small-scale FFTs whose data can fit in the local storage of a single accelerator. Real-world FFT applications often require much larger scale FFTs, but it is extremely challenging for heterogeneous multi-core system with distributed architectures to make efficient large FFT implementations. In this paper, we introduce the first known FFT library for the heterogeneous multi-core system with distributed architecture that can solve one-dimensional FFTs larger than what fits in a single accelerator. Our implementation achieves 67% performance improvement of FFTW 3.2.2 (Fastest Fourier Transform in the West) and sustains over 36 single precision FFT GFLOPs ‘end-to-end’ Achieving such high performance requires novel schemes for large-scale FFT factorization, data permutation and all-to-all exchanges, and buffer designs to maximize use of the local storage while minimizing communication overhead. One important finding in this paper is that large-scale FFT on this kind of architecture behaves as data transfer bound which is quite different from other architectures. A significant contribution of this paper is that for each major component of our algorithm, we explore many possible design options and present quantitative performance comparisons. This provides value beyond specific architecture, as it illustrates the fundamental features associated with different communication paradigms and mechanisms for the heterogeneous multi-core system. Today’s computer systems are increasingly being designed to include general purpose accelerators. The techniques in this paper can also be applied to these architectures especially when they have limited local storage or steep cache hierarchies. We also provide insights on applying techniques in this paper to similar architectures.

Key advances in the presilicon functional verification of the IBM zEnterprise microprocessor and storage hierarchy

This paper highlights key advances in the presilicon verification effort of the IBM zEnterprise® 196 (z196) microprocessor and storage hierarchy. It focuses on the unique set of verification challenges as well as the process innovations that address them. At the time of product launch, the z196 system represented the industry's fastest and most scalable enterprise system, with up to 80 customer-configurable out-of-order core processors operating at 5.2 GHz. In addition to offering industry-leading performance, the z196 system builds upon its leadership in reliability by introducing a new redundant array of independent memory (RAIM) technology into its memory subsystem. The new product features in this system drove innovations in all aspects of processor functional verification, including stimulus generation, functional checking, debugging, and coverage. A new hybrid RAIM verification methodology, which includes both formal and random methods, is described. Many process and methodology improvements were made to improve developmental collaboration across a global team. These enhancements include a simulation development environment that uses common shared components across functional partitions, as well as a shared cache loader that was used across multiple environments. We also present a self-configuring test-case generation process that focused on the coverage of functional stimulus.

Where does Hydrogen Fit in a Sustainable Energy Economy?

Abstract Where does hydrogen fit into a global sustainable energy strategy for the 21 st century, as we face the enormous challenges of irreversible climate change and uncertain oil supply? This fundamental question is addressed by sketching a sustainable energy strategy that is based predominantly on renewable energy inputs and energy efficiency, with hydrogen playing a crucial and substantial role. But this role is not an exclusive one as in the original concept of the ‘hydrogen economy’ proposed in the early 1970s. A hierarchy of spatially-distributed hydrogen production, storage and distribution centres relying on local renewable energy sources and feedstocks would be created to avoid the need for an expensive long-distance hydrogen pipeline system. There would thus be complementary use of electricity and hydrogen as energy vectors. Importantly, bulk hydrogen storage would provide the strategic energy reserve to guarantee national and global energy security in a world relying increasingly on renewable energy; and longer-term seasonal storage on electricity grids relying mainly on renewables. In the transport sector, a ‘horses for courses’ approach is proposed in which hydrogen fuel cell vehicles would be used in road and rail vehicles requiring a range comparable to today's petrol and diesel vehicles, and in coastal and international shipping, while liquid hydrogen would probably have to be used in air transport. Plug-in battery electric vehicles would be reserved for shorter-trips. Energy-economic-environmental modelling is recommended as the next step to quantify the net benefits of the overall strategy outlined.

Re-envisioning the role of hydrogen in a sustainable energy economy

This paper addresses the fundamental question of where hydrogen might fit into a global sustainable energy strategy for the 21st century that confronts the three-pronged challenge of irreversible climate change, uncertain oil supply, and rising pollution. We re-envision the role of hydrogen at national and international strategic levels, relying entirely on renewable energy and energy efficiency. It is suggested the time for an exclusive ‘hydrogen economy’ has passed, since electricity and batteries would be used extensively as well. Yet hydrogen would still play a crucial role: in road and rail vehicles requiring a range comparable to today’s petrol and diesel vehicles; in coastal and international shipping; in air transport; and for longer-term seasonal storage on electricity grids relying mainly on renewables. Hydrogen fuel cell vehicles are proposed where medium and long distance trips are required, with plug-in battery electric vehicles reserved for just short trips. A hierarchy of spatially-distributed hydrogen production, storage and distribution centers relying on local renewable energy sources and feedstocks would be created to limit the required hydrogen pipeline network to the main metropolitan areas and regions by complementary use of electricity as a major energy vector. Bulk hydrogen storage would provide the strategic energy reserve to guarantee national and global energy security in a world relying increasingly on renewable energy. It is recommended that this vision next be applied to specific countries by conducting detailed energy-economic-environmental modeling to quantify its net benefits.