An Active Storage System for High Performance Computing

In addition to the performance, power consumption has become a major concern in High Performance Computing (HPC) systems. Typically the cooling system and the IT loads are the major contributors to the power bills. Understanding the power consumption at different granular levels in the HPC system is a first step to quantify the problems in HPC system. By proper monitoring and effective utilization of the cooling system, the power requirements of the HPC facility can be effectively met. In this paper we present a system where node level power measurement and WSN based rack level temperature measurement are used to provide localized control of cold air supply. A Smart Power Monitoring and Distribution Unit (PMDU) is designed and developed, to replace the existing Power Distribution Unit (PDU) in HPC. This can measure and report the power consumption, to support power profiling of large scale HPC system. This measured data are communicated to a base station via Ethernet. This base station collects all such measurements which can be used for power profiling of IT load of the HPC system. This helps to provide better insight into the power utilization pattern. Wireless Sensor Network (WSN) is used to collect exhaust and inlet air temperature of the server nodes and this information is used for directing the cold air effectively. A vent control system is designed and fabricated for intelligently directing the air flow to the server node inlet. It takes the node power and the temperature data as its inputs. This enables supplying/ redirecting more cold air towards under-cooled nodes without creating an extra load on the cooling system, thereby bringing in effective cooling at reduced power consumption. As an added advantage this could help in hot-spots mitigation.

With the explosion of data, we have an urgent demand for data throughput in high performance computing systems. Data-intensive applications are becoming increasingly common in HPC environments. As data scale increases faster than systems, it’s time to fully utilize resources in every aspect, including computing power, storage capacity and data throughput. We can no longer ignore data preprocessing since it’s an important procedure, especially when dealing with large amount of data. How to efficiently perform data preprocessing in current HPC systems? How to make full use of system resources on data-intensive applications? What should be valued when designing new HPC architectures? All these questions need answers. In this paper, we drew a sketch for procedure of data-intensive applications, which lead to an adaptive resource allocation scheme according to procedure requirements. We analyzed characters of preprocessing and designed a preprocessing model for data-intensive applications in HPC systems. It has not only fulfilled the demand for computing but also meet the need of throughput, with cooperative work in storage system and storage management system. Experiments were done on Sunway TaihuLight, one of the world’s fastest supercomputers. The whole procedure of preprocessing at Petabytes can be done in hours without interfering other ongoing applications.

Scientific workflows are increasingly transferring large amounts of data between high performance computing (HPC) systems. Even though these HPC systems are connected via high-speed dedicated networks and use dedicated data transfer nodes (DTNs), it is still difficult to predict the data transfer throughput because of variations in data transfer protocols, host configurations, performance of file systems, and overlapping workloads. In order to provide reliable performance prediction for better resource management and job scheduling, we need models for predicting data transfer throughput under real-world conditions. In this paper, we explore different machine learning approaches for building data-driven models to improve performance and prediction of large-scale data transfer throughput. In addition to the variables already collected by the network monitoring system, we also develop heuristics to derive additional metrics for improving the prediction accuracy. We use the prediction results to identify the importance of different network parameters in predicting the throughput for large-scale data transfers. Through extensive tests, we identify key network parameters, discover interesting variations among different HPC sites, and show that we can predict throughput with high accuracy. We also analyze our models and results to provide recommendations for improving the performance of big data transfers.

Large-scale storage systems are crucial components in data-intensive applications such as search engine clusters, video-on-demand servers, sensor networks and grid computing. A storage server typically consists of a set of storage devices. In such systems, data layouts may need to be reconfigured over time for load balancing or in the event of system failure/upgrades. It is critical to migrate data to their target locations as quickly as possible to obtain the best performance. Most of the previous results on data migration assume that each storage node can perform only one data transfer at a time. A storage node, however, can typically handle multiple transfers simultaneously and this can reduce the total migration time significantly. Moreover, storage devices tend to have heterogeneous capabilities as devices may be added over time due to storage demand increase. In this paper, we consider the heterogeneous data migration problem, where we assume that each storage node v has different transfer constraint cv, which represents how many simultaneous transfers v can handle. We develop algorithms to minimize the data migration time. We show that it is possible to find an optimal migration schedule when all cvs are even. Furthermore, though the problem is NP-hard in general, we give an efficient algorithm that offers a rigorous (1 + o(1))-approximation guarantee.

With the rapid development of enterprise information construction, the amount of business data shows a blowout growth. The traditional storage architecture in data storage, real-time migration, dynamic expansion, and other aspects cannot meet the current development needs. Ceph is a reliable, auto-rebalance, auto-recovery distributed storage system that supports thousands of storage nodes. Ceph nodes have a large amount of data and information communication, involving data backup, node redundancy, and a large number of user read and write functions, so the monitoring and prediction of network traffic can make more efficient use of the network. The traditional traffic scheduling problem will limit the scheduling performance of the Ceph platform. Therefore, this paper mainly adopts Software Defined Network (SDN) to solve the platform scheduling problem, uses SDN to collect traffic data between storage nodes of Ceph, and uses Long Short-Term Memory (LSTM) to predict Network traffic. It then combines with the Dijkstra algorithm to achieve load balancing of the data plane. Experiments show that the proposed scheme can monitor and predict network traffic and optimize the load balance of storage platform traffic.

We have proposed a distributed storage system which aims to utilize storage fragments distributed in network nodes without administrators' maintenance cost. Even the performance of storage devices are not uniform, autonomously data block migration helps to achieve enhancing the whole I/O performance of the storage system. However, we have considered all the storage devices to be reliable equally. On this assumption, migration of an important data block to unstable storage might cause degradation of the availability of the data block. In this paper, we propose a scheme to choose the destination of data block migration by evaluating the nodes' reliability along with the performance of storage devices. This scheme aims to improve system stability by preventing to locate a critical data onto an unstable storage device. The storage nodes which lack stability are also utilized for storing a data block which can be easily regenerated or duplicated. The results of the preliminary investigation show that a large number of warning logs from Machine Check Architecture may suggest a sign of unexpected system shutdown.

Secure communication over public networks like the Internet requires the use of cryptographic algorithms as basic building blocks. Most cryptographic workloads pose a considerable burden on devices like PDAs, cell phones, and sensor nodes, which are limited in processing power, memory and energy. In this paper we present an approach to increase the efficiency of 32-bit processors for handling symmetric cryptographic algorithms with the help of instruction set extensions. We propose a number of custom instructions to support the Advanced Encryption Standard (AES). Using the SPARC V8-compatible Leon2 embedded processor, we evaluate the effects of the extensions on performance and code size of AES, as well as on silicon area. With a moderate increase in silicon area, AES performance can be improved by a factor of nearly 10, while code size is reduced significantly and implementation flexibility is retained. We also show that our approach is very beneficial for implementation in superscalar processors and that it can compete with the performance of previously proposed cryptographic processors and instruction set extensions.

Scientific discovery is nowadays driven by large-scale simulations running on massively parallel high-performance computing (HPC) systems. These applications each generate a large amount of data, which then needs to be post-processed for example for data mining or visualization. Unfortunately, the computing platform used for post processing might be different from the one on which the data is initially generated, introducing the challenge of moving large amount of data between computing platforms. This is especially challenging when these two platforms are geographically separated since the data needs to be moved between computing facilities. This is even more critical when scientists tightly couple their domain specific applications with a post processing application.The paper presents a solution for the data transfer between MPI applications using a dedicated wide area network (WAN) terabit infrastructure. The proposed solution is based on parallel access to data files and the Message Passing Interface (MPI) over the Common Communication Infrastructure (CCI) for the data transfer over a routed infrastructure. In the context of this research, the Energy Sciences Network (ESnet) of the U.S. Department of Energy (DOE) is targeted for the transfer of data between DOE national laboratories.

All the machines are required to be under a common administrator and be able to communicate securely. To communicate securely, Advanced Encryption Standard (AES) algorithm is used to enable protection of data at each clusters. It performs encryption and decryption before read and write respectively. Encryption and Decryption in key is used for securing Hadoop Distributed File System. The existing system depends on a single name node to manage almost all operations of every data block in the file system. As a result it can be a bottleneck resource and a single point of failure. To overcome this failure, Load Rebalancing Algorithm is used for Hadoop Distributed File System using Distributed Hash Table. The Proposed load rebalancing algorithm will be compared against a centralized approach in a production system and a competing distributed solution is presented in the literature. The storage nodes are structured as a network based on distributed hash table discovering a file chunk can simply refer to rapid key lookup in DHTs, given that a unique handle (or identifier) is assigned to each file chunk.

Chapter 6 - Secondary Storage

Simulation environments provide a comprehensive set of advantages to users, like cost effectiveness and capability to understand the shortcomings of the system under design, without physical implementation. Simulation platforms help the industry and academia, to document and publish their research outputs in a timely and cost efficient manner. The ECSim tool presented here, is meant for academic use in the initial stages of research. The platform provides an environment where the performance of erasure coded storage systems can be tested without much effort. The main highlight of the simulator is that it provides a very simple environment which can run on a standalone system. The environment does not require the user to be a programmer, since it provides an interactive command line interface to the user. The ability to simulate data center, clusters, master nodes, storage nodes with computing power, storage devices, bandwidth usage and disk I/O involved are notable features of ECSim.

Network storages are used widely in many fields of our society. But their testing performances are not as expected. Most of the tools being used to test the performance of network storage comes from the tools used to test the traditional storage system because the storage devices connected to the net act as the local storage devices. So these tools inevitably miss the effects exerted by the networks which connect the hosts to the storage devices. For example, these tools can hardly generate requests which exceed the maximum loading of storage devices. In this paper, we propose one new method to test the performance of network storage which can easily generate requests that exceed the maximum loading of storage devices. In this method, the test program sends the requests by fixed frequency and the initiator will not be restrained by the targeter when the quantity of requests is close to the maximum load. The load simulated by the new method is more like the load in the real world, especially when the load is very high.

Storage density of the traditional cloud storage system is not big and the comprehensive storage efficiency is low. When the large amount of data increases dramatically, the storage efficiency of the system would be decreased seriously. Therefore, with the development of cloud storage technology, an application of Markov Decision Process in Scheduling Strategy on cloud storage (MDPSS) is proposed in this paper. MDPSS can predict periodically in stages by using the steady Markov decision process and select the optimal storage node easily. By scheduling the data to the optimal node in real time, MDPSS can enhance the flexibility of the data storage and reduce the storage cost. Simulation results show that MDPSS has the ability to predict accurately when the size of the data is different.

The solution of large eigenproblems is involved in many scientific and engineering applications when for instance, stability analysis is a concern. For large simulation in material physics or thermo-acoustics, the calculation can last for many hours on large parallel platforms. On future large-scale systems, the mean time between failures (MTBF) of the system is expected to decrease so that many faults could occur during the solution of large eigenproblems. Consequently, it becomes critical to design parallel eigensolvers that can survive faults. In that framework, we investigate the relevance of approaches relying on numerical techniques, which might be combined with more classical techniques for real large-scale parallel implementations. Because we focus on numerical remedies we do not consider parallel implementations nor parallel experiments but only numerical experiments. We assume that a separate mechanism ensures the fault detection and that a system layer provides support for setting back the environment (processes,. . .) in a running state. Once the system is in a running state, after a fault, our main objective is to provide robust resilient schemes so that the eigensolver may keep converging in the presence of the fault without restarting the calculation from scratch. For this purpose, we extend the interpolation-restart (IR) strategies initially introduced for the solution of linear systems in a previous work to the solution of eigenproblems in this paper. For a given numerical scheme, the IR strategies consist of extracting relevant spectral information from available data after a fault. After data extraction, a well-selected part of the missing data is regenerated through interpolation strategies to constitute a meaningful input to restart the numerical algorithm. One of the main features of this numerical remedy is that it does not require extra resources, i.e., computational unit or computing time, when no fault occurs. In this paper, we revisit a few state-of-the-art methods for solving large sparse eigenvalue problems namely the Arnoldi methods, subspace iteration methods and the Jacobi-Davidson method, in the light of our IR strategies. For each considered eigensolver, we adapt the IR strategies to regenerate as much spectral information as possible. Through extensive numerical experiments, we study the respective robustness of the resulting resilient schemes with respect to the MTBF and to the amount of data loss via qualitative and quantitative illustrations. 1. Introduction. The computation of eigenpairs (eigenvalues and eigenvectors) of large sparse matrices is involved in many scientific and engineering applications such as when stability analysis is a concern. To name a few, it appears in structural dynamics, thermodynamics, thermo-acoustics, quantum chemistry. With the permanent increase of the computational power of high performance computing (HPC) systems by using a larger and larger number of CPU cores or specialized processing units, HPC applications are increasingly prone to faults. To guarantee fault tolerance, two classes of strategies are required. One for the fault detection and the other for fault correction. Faults such as computational node crashes are obvious to detect while silent faults may be challenging to detect. To cope with silent faults, a duplication strategy is commonly used for fault detection [18, 39] by comparing the outputs, while triple modular redundancy (TMR) is used for fault detection and correction [34, 37]. However, the additional computational resources required by such replication strategies may represent a severe penalty. Instead of replicating computational resources, studies [7, 36] propose a time redundancy model for fault detection. It consists in repeating computation twice on the same resource. The advantage of time redundancy models is the flexibility at application level; software developers can indeed select only a set of critical instructions to protect. Recomputing only some instructions instead of the whole application lowers the time redundancy overhead [25]. In some numerical simulations, data naturally satisfy well defined mathematical properties. These properties can be efficiently exploited for fault detection through a periodical check of the numerical properties during computation [10]. Checkpoint/restart is the most studied fault recovery strategy in the context of HPC systems. The common checkpoint/restart scheme consists in periodically saving data onto a reliable storage device such as a remote disk. When a fault occurs, a rollback is performed to the point of the most recent and consistent checkpoint. According to the implemented checkpoint strategy, all processes

The COVID-19 pandemic brought Bioinformatics into the spotlight, revealing that several existing methods, algorithms, and tools were not well prepared to handle large amounts of genomic data efficiently. This led to prohibitively long execution times and the need to reduce the extent of analyses to obtain results in a reasonable amount of time. In this survey, we review available high-performance computing and hardware-accelerated systems based on FPGA and GPU technology. Optimized and hardware-accelerated systems can conduct more thorough analyses considerably faster than pure software implementations, allowing to reach important conclusions in a timely manner to drive scientific discoveries. We discuss the reasons that are currently hindering high-performance solutions from being widely deployed in real-world biological analyses and describe a research direction that can pave the way to enable this.