More Like this

1. Expanding an HPC Cluster to Support the Computational Demands of Digital Pathology

   https://doi.org/10.1109/SPMB.2018.8615614  

   Campbell, C. ; Mecca, N. ; Duong, T. ; Obeid, I. ; Picone, J. ( December 2018 , IEEE Signal Processing in Medicine and Biology Symposium (SPMB))

   Obeid, Iyad ; Selesnick, Ivan ; Picone, Joseph (Ed.)

   The goal of this work was to design a low-cost computing facility that can support the development of an open source digital pathology corpus containing 1M images [1]. A single image from a clinical-grade digital pathology scanner can range in size from hundreds of megabytes to five gigabytes. A 1M image database requires over a petabyte (PB) of disk space. To do meaningful work in this problem space requires a significant allocation of computing resources. The improvements and expansions to our HPC (highperformance computing) cluster, known as Neuronix [2], required to support working with digital pathology fall into two broad categories: computation and storage. To handle the increased computational burden and increase job throughput, we are using Slurm [3] as our scheduler and resource manager. For storage, we have designed and implemented a multi-layer filesystem architecture to distribute a filesystem across multiple machines. These enhancements, which are entirely based on open source software, have extended the capabilities of our cluster and increased its cost-effectiveness. Slurm has numerous features that allow it to generalize to a number of different scenarios. Among the most notable is its support for GPU (graphics processing unit) scheduling. GPUs can offer a tremendous performance increase in machine learning applications [4] and Slurm’s built-in mechanisms for handling them was a key factor in making this choice. Slurm has a general resource (GRES) mechanism that can be used to configure and enable support for resources beyond the ones provided by the traditional HPC scheduler (e.g. memory, wall-clock time), and GPUs are among the GRES types that can be supported by Slurm [5]. In addition to being able to track resources, Slurm does strict enforcement of resource allocation. This becomes very important as the computational demands of the jobs increase, so that they have all the resources they need, and that they don’t take resources from other jobs. It is a common practice among GPU-enabled frameworks to query the CUDA runtime library/drivers and iterate over the list of GPUs, attempting to establish a context on all of them. Slurm is able to affect the hardware discovery process of these jobs, which enables a number of these jobs to run alongside each other, even if the GPUs are in exclusive-process mode. To store large quantities of digital pathology slides, we developed a robust, extensible distributed storage solution. We utilized a number of open source tools to create a single filesystem, which can be mounted by any machine on the network. At the lowest layer of abstraction are the hard drives, which were split into 4 60-disk chassis, using 8TB drives. To support these disks, we have two server units, each equipped with Intel Xeon CPUs and 128GB of RAM. At the filesystem level, we have implemented a multi-layer solution that: (1) connects the disks together into a single filesystem/mountpoint using the ZFS (Zettabyte File System) [6], and (2) connects filesystems on multiple machines together to form a single mountpoint using Gluster [7]. ZFS, initially developed by Sun Microsystems, provides disk-level awareness and a filesystem which takes advantage of that awareness to provide fault tolerance. At the filesystem level, ZFS protects against data corruption and the infamous RAID write-hole bug by implementing a journaling scheme (the ZFS intent log, or ZIL) and copy-on-write functionality. Each machine (1 controller + 2 disk chassis) has its own separate ZFS filesystem. Gluster, essentially a meta-filesystem, takes each of these, and provides the means to connect them together over the network and using distributed (similar to RAID 0 but without striping individual files), and mirrored (similar to RAID 1) configurations [8]. By implementing these improvements, it has been possible to expand the storage and computational power of the Neuronix cluster arbitrarily to support the most computationally-intensive endeavors by scaling horizontally. We have greatly improved the scalability of the cluster while maintaining its excellent price/performance ratio [1]. 

   more » « less
2. Append is Near: Log-based Data Management on ZNS SSDs

   Purandare, Devashish ; Wilcox, Pete ; Litz, Heiner ; Finkelstein, Shel ( January 2022 , 12th Annual Conference on Innovative Data Systems Research (CIDR ’22).)

   Log-based data management systems use storage as if it were an append-only medium, transforming random writes into sequential writes, which delivers significant benefits when logs are persisted on hard disks. Although solid-state drives (SSDs) offer improved random write capabilities, sequential writes continue to be advan- tageous due to locality and space efficiency. However, the inherent properties of flash-based SSDs induce major disadvantages when used with a random write block interface, causing write amplifica- tion, uneven wear, log stacking, and garbage collection overheads. To eliminate these disadvantages, Zoned Namespace (ZNS) SSDs have recently been introduced. They offer increased capacity, re- duced write amplification, and open up data placement and garbage collection to the host through zones, which have sequential-write semantics and must be explicitly reset. We explain how the new ZNS Zone Append primitive, which sup- ports pushing fine-grained data placement onto the device, along with our proposal for “Group Append”, which enables sub-block sized appends, could benefit log-structured data management sys- tems. We explore advantages of ZNS SSDs with Zone Append, Group Append, and computational storage in four log-based data management areas: (i) log-based file systems, (ii) LSM trees such as RocksDB, (iii) database systems, and (iv) event logs/shared logs. Furthermore, we propose research directions for each of these data management systems using ZNS SSDs. 

   more » « less
3. ConEx: Efficient Exploration of Big-Data System Configurations for Better Performance

   https://doi.org/10.1109/TSE.2020.3007560  

   Krishna, Rahul ; Tang, Chong ; Sullivan, Kevin ; Ray, Baishakhi ( July 2020 , IEEE Transactions on Software Engineering)

   Configuration space complexity makes the big-data software systems hard to configure well. Consider Hadoop, with over nine hundred parameters, developers often just use the default configurations provided with Hadoop distributions. The opportunity costs in lost performance are significant. Popular learning-based approaches to auto-tune software does not scale well for big-data systems because of the high cost of collecting training data. We present a new method based on a combination of Evolutionary Markov Chain Monte Carlo (EMCMC)} sampling and cost reduction techniques tofind better-performing configurations for big data systems. For cost reduction, we developed and experimentally tested and validated two approaches: using scaled-up big data jobs as proxies for the objective function for larger jobs and using a dynamic job similarity measure to infer that results obtained for one kind of big data problem will work well for similar problems. Our experimental results suggest that our approach promises to improve the performance of big data systems significantly and that it outperforms competing approaches based on random sampling, basic genetic algorithms (GA), and predictive model learning. Our experimental results support the conclusion that our approach strongly demonstrates the potential toimprove the performance of big data systems significantly and frugally. 

   more » « less
4. Henosis: workload-driven small array consolidation and placement for HDF5 applications on heterogeneous data stores

   https://doi.org/10.1145/3330345.3330380  

   Kang, Donghe ; Patel, Vedang ; Nair, Ashwati ; Blanas, Spyros ; Wang, Yang ; Parthasarathy, Srinivasan ( June 2019 , Proceedings of the 33rd ACM International Conference on Supercomputing - ICS '19)

   Scientific data analysis pipelines face scalability bottlenecks when processing massive datasets that consist of millions of small files. Such datasets commonly arise in domains as diverse as detecting supernovae and post-processing computational fluid dynamics simulations. Furthermore, applications often use inference frameworks such as TensorFlow and PyTorch whose naive I/O methods exacerbate I/O bottlenecks. One solution is to use scientific file formats, such as HDF5 and FITS, to organize small arrays in one big file. However, storing everything in one file does not fully leverage the heterogeneous data storage capabilities of modern clusters. This paper presents Henosis, a system that intercepts data accesses inside the HDF5 library and transparently redirects I/O to the in-memory Redis object store or the disk-based TileDB array store. During this process, Henosis consolidates small arrays into bigger chunks and intelligently places them in data stores. A critical research aspect of Henosis is that it formulates object consolidation and data placement as a single optimization problem. Henosis carefully constructs a graph to capture the I/O activity of a workload and produces an initial solution to the optimization problem using graph partitioning. Henosis then refines the solution using a hill-climbing algorithm which migrates arrays between data stores to minimize I/O cost. The evaluation on two real scientific data analysis pipelines shows that consolidation with Henosis makes I/O 300× faster than directly reading small arrays from TileDB and 3.5× faster than workload-oblivious consolidation methods. Moreover, jointly optimizing consolidation and placement in Henosis makes I/O 1.7× faster than strategies that perform consolidation and placement independently. 

   more » « less
5. Data Recovery for 3D Magnetic Recording

   Kheong Chan, Ahmed Aboutaleb ( July 2019 , 2019 Magnetic Recording Conference (TMRC 2019))

   Solid State Drives (SSD) compete with Hard Disk Drives (HDD) in the data storage market. Recent advances in SSD capacity/cost have come from arranging the flash memory cells not just on the 2D surface but from also stacking many cells vertically through the 3rd dimension. The same option has not been seen as a practical approach for HDD technology that is based on magnetic recording. Data can only be written to and read from just above the surface of the medium, and any data on additional layers deeper in the medium is profoundly affected by the additional spacing and loss of resolution. Nevertheless, modest gains may be still be possible. Earlier work suggested gains around 17% for two stacked layers. That work only examined a single isolated track on each of two layers and just one reader. In this new work, we examine a minimal 3D configuration again comprising two layers, where two adjacent tracks on the upper layer straddle a double width track on the lower layer. We take the writing process as a given—for instance utilizing Microwave Assisted Magnetic Recording. For readback, we variously assume 1, 2, or 3 readers arrayed above the data tracks. 

   more » « less