{"id":23312,"date":"2019-09-23T08:30:57","date_gmt":"2019-09-23T15:30:57","guid":{"rendered":"https:\/\/insidebigdata.com\/?p=23312"},"modified":"2019-10-07T11:51:23","modified_gmt":"2019-10-07T18:51:23","slug":"daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement","status":"publish","type":"post","link":"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/","title":{"rendered":"DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement"},"content":{"rendered":"\n<p>Forget what you previously knew about high-performance storage and file systems. New I\/O models for HPC such as Distributed Asynchronous Object Storage (<a href=\"https:\/\/software.intel.com\/en-us\/videos\/daos-high-performance-software-defined-storage-over-persistent-memory\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\">DAOS<\/a>) have been architected from the ground up to make use of new NVM technologies such as Intel\u00ae Optane\u2122 DC Persistent Memory Modules (Intel Optane DCPMMs). With latencies measured in nanoseconds and bandwidth measured in tens of GB\/s, new storage devices such as Intel DCPMMs redefine the measures used to describe high-performance nonvolatile storage.<\/p>\n\n\n\n<blockquote class=\"wp-block-quote\"><p>DAOS is an extremely lightweight I\/O infrastructure that operates end-to-end in user space with full operating system bypass to deliver the performance capability of sub-microsecond storage devices. DAOS offers a shift away from the traditional HPC block-based, high-latency <a rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\" href=\"https:\/\/en.wikipedia.org\/wiki\/POSIX\" target=\"_blank\">POSIX<\/a> storage model to one that inherently supports fine-grained data access and unlocks the performance of next generation storage systems.<\/p><\/blockquote>\n\n\n\n<p>The Argonne Leadership Computing Facility will be the first major production deployment of the DAOS storage system as part of Aurora, the first US exascale system coming in 2021. The DAOS storage system is designed to provide the levels of metadata operation rates and bandwidth required for I\/O extensive workloads on an exascale-level machine. [1]<\/p>\n\n\n\n<p><strong>Designed to bypass the traditional POSIX IO bottlenecks<\/strong><\/p>\n\n\n\n<p>The key to DAOS performance when used in combination with\nthe Intel Optane DCPMM module is that the CPU can communicate in user space directly\nwith storage via special Direct Access (DAX) instructions available\non&nbsp;second generation Intel\u00ae Xeon\u00ae Scalable Family processors.<\/p>\n\n\n\n<p>The difference is significant as illustrated in Figure 2, because direct user space access eliminates many of the frustrating performance limiting aspects of traditional POSIX I\/O that increase latency and reduce storage bandwidth such as the operating system, page cache, block driver, and PCIe bus. <\/p>\n\n\n\n<blockquote class=\"wp-block-quote\"><p>The Argonne Leadership Computing Facility will be the first major production deployment of the DAOS storage system as part of Aurora, the first US exascale system coming in 2021. The DAOS storage system is designed to provide the levels of metadata operation rates and bandwidth required for I\/O extensive workloads on an exascale-level machine.  <\/p><\/blockquote>\n\n\n\n<p><strong>Designed to bypass the traditional POSIX IO bottlenecks<\/strong><\/p>\n\n\n\n<p>The key to DAOS performance when used in combination with the Intel Optane DCPMM module is that the CPU can communicate in user space directly with storage via special Direct Access (DAX) instructions available on&nbsp;second generation Intel\u00ae Xeon\u00ae Scalable Family processors. <\/p>\n\n\n\n<p>The difference is significant as illustrated in Figure 2, because direct user space access eliminates many of the frustrating performance limiting aspects of traditional POSIX I\/O that increase latency and reduce storage bandwidth such as the operating system, page cache, block driver, and PCIe bus.  <\/p>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter\"><img decoding=\"async\" loading=\"lazy\" width=\"700\" height=\"363\" src=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/New-DAOS-architecture.png\" alt=\"\" class=\"wp-image-23352\" srcset=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/New-DAOS-architecture.png 700w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/New-DAOS-architecture-150x78.png 150w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/New-DAOS-architecture-300x156.png 300w\" sizes=\"(max-width: 700px) 100vw, 700px\" \/><\/figure><\/div>\n\n\n\n<p>Overall, the DAOS API and data model are driving a new scalable storage model for structured and unstructured data sets that supports a rich set of data models that can capitalize on the performance of Intel Optane non-volatile memory. DAOS supports both non-blocking data and metadata operations are supported plus a burgeoning software ecosystem already includes a number of rich data models. <\/p>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter\"><img decoding=\"async\" loading=\"lazy\" width=\"1024\" height=\"229\" src=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS2-1024x229.png\" alt=\"\" class=\"wp-image-23314\" srcset=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS2-1024x229.png 1024w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS2-150x34.png 150w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS2-300x67.png 300w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS2-768x172.png 768w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS2.png 2004w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><figcaption>Figure 2: DAOS is a non-POSIX rich storage API that provides a new foundation for HPC I\/O<\/figcaption><\/figure><\/div>\n\n\n\n<p><strong>DAOS is open-source, available now<\/strong><\/p>\n\n\n\n<p>DAOS is available now. Simply download from the DAOS <a href=\"https:\/\/github.com\/daos-stack\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\">github<\/a> repository and install to evaluate the benefits in your environment. Numerous paths can be followed to test and integrate DAOS into your HPC application environment:<\/p>\n\n\n\n<ul><li>A high-performance MPICH and <a href=\"https:\/\/www.mcs.anl.gov\/projects\/romio\/\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\"ROMIO (opens in a new tab)\">ROMIO<\/a> middleware layer is provided which means many HPC codes will realize full DAOS performance immediately without code modification.<\/li><li>Similarly, <a href=\"https:\/\/github.com\/daos-stack\/daos\/blob\/master\/src\/include\/daos_fs.h\">libdfs<\/a>&nbsp;implements files and directories over the DAOS API by encapsulating a POSIX namespace into a DAOS container. This library can be linked directly with the application or mounted locally through <a href=\"https:\/\/en.wikipedia.org\/wiki\/Filesystem_in_Userspace\">FUSE<\/a>. <\/li><li>Native C, Python, and Go interfaces are provided, which means most HPC applications can be ported to directly utilize DAOS.<\/li><li>Other popular IO middleware layers have been integrated into DAOS including HDFS for Hadoop codes.<\/li><li>Persistent Memory Over Fabric (PMOF) enables replication of data remotely between machines with persistent memory.<\/li><li>For those in a rush, the DAOS software can be tested without persistence using DRAM via tmpfs.  <\/li><\/ul>\n\n\n\n<p><strong>Intel\u00ae Optane\u2122 DCPMM technology redefines storage performance \u2013 even without DAOS<\/strong> <\/p>\n\n\n\n<p>By ingeniously packaging Intel Optane memory in a DDR form factor so it can sit on the processor\u2019s memory bus, Intel Optane DCPMM literally requires new units of measure be applied to the performance of non-volatile storage. When properly optimized, Intel Optane DCPMM devices can provide up to multiple tens of GB\/s of throughput at nanosecond latencies versus single digit GB\/s of throughput at microsecond latencies seen with high performing NAND-based flash storage devices. [2] Even without DAOS, Intel Optane DCPMM can make key storage applications up to 17 times faster. [3]  To do so, simply use the Intel <a href=\"https:\/\/software.intel.com\/en-us\/articles\/introduction-to-programming-with-persistent-memory-from-intel\">Persistent Memory Development Kit (PMDK)<\/a> to support the transactional operations necessary to keep the persistent data consistent and durable, or run with the XFS, EXT4, or NTFS file systems optimized to use Intel\u00ae Optane\u2122 persistent storage. <\/p>\n\n\n\n<p>Along with storage, Intel Optane DCPMM memory can be used as a non-persistent memory tier to augment main memory. This means that \u201cfat\u201d computational nodes can essentially contain many terabytes of main memory that delivers performance comparable to DRAM for many real world applications, [4] which can help with HPC simulations that struggle with memory capacity limitations on conventional DRAM-only systems, and without needing to introduce any changes to the application. [5] <\/p>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter\"><img decoding=\"async\" loading=\"lazy\" width=\"1024\" height=\"487\" src=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS3-1024x487.png\" alt=\"\" class=\"wp-image-23315\" srcset=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS3-1024x487.png 1024w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS3-150x71.png 150w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS3-300x143.png 300w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS3-768x365.png 768w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/DAOS3.png 1935w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><figcaption>Figure 3: There are many form factors and ways to use Intel\u00ae Optane\u2122 memory<\/figcaption><\/figure><\/div>\n\n\n\n<p>Overall, a burgeoning software ecosystem is rapidly evolving that shows what is possible for many HPC applications including those that use extremely large, globally accessible data sets. <a rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\" href=\"https:\/\/www.intel.com\/content\/www\/us\/en\/software\/apache-spark-optimization-technology-brief.html\" target=\"_blank\">Intel optimized Apache Spark, for example, doubles throughput and reduces runtime by up to 40%<\/a> for data centric workloads. Other partner frameworks such as <a rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\" href=\"https:\/\/www.intel.com\/content\/www\/us\/en\/big-data\/partners\/sap\/overview.html\" target=\"_blank\">SAP HANA<\/a>, <a rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\" href=\"https:\/\/software.intel.com\/en-us\/articles\/intel-optane-dc-persistent-memory-for-greater-capacity-affordability-persistence\" target=\"_blank\">Redis Labs<\/a>, <a rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\" href=\"https:\/\/www.intel.com\/content\/www\/us\/en\/processors\/xeon\/scalable\/software-solutions\/asiainfo-comms-telco-bss-brief.html\" target=\"_blank\">AsiaInfo<\/a>, <a rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\" href=\"https:\/\/www.intel.com\/content\/www\/us\/en\/processors\/xeon\/scalable\/software-solutions\/aerospike-enterprise-database-brief.html\" target=\"_blank\">Aerospike<\/a>, and a number of <a href=\"https:\/\/www.intel.com\/content\/www\/us\/en\/processors\/xeon\/scalable\/software-solutions\/overview.html\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\">ISVs<\/a> demonstrate how the bottleneck imposed by the current popular HPC parallel distributed file-systems such as Lustre and GPFS can be eliminated.  <\/p>\n\n\n\n<p>In terms of raw performance, the graphic below shows that memory-system based Intel DCPMM persistent memory can be accessed in nanoseconds. In comparison, it takes over 80 \u00b5s for a modern NAND-based SSD, which plugs into the PCIe bus and communicates using the NVM Express protocol, to read a single block of data. <\/p>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter\"><img decoding=\"async\" loading=\"lazy\" width=\"1024\" height=\"540\" src=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/Fig-4_v2-01-1024x540.png\" alt=\"\" class=\"wp-image-23316\" srcset=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/Fig-4_v2-01-1024x540.png 1024w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/Fig-4_v2-01-150x79.png 150w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/Fig-4_v2-01-300x158.png 300w, https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/09\/Fig-4_v2-01-768x405.png 768w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><figcaption>Figure 4: Extreme performance with Intel\u00ae Optane\u2122 DC Persistent Memory<\/figcaption><\/figure><\/div>\n\n\n\n<p><strong>Summary<\/strong><\/p>\n\n\n\n<p>Forget what you previously knew about high-performance storage and file systems. In combination, DAOS plus Intel Optane DCPMM provide a more performant (i.e. lower latency, higher BW\/IOPS), scalable (i.e. millions of MPI tasks and thousands of storage servers) plus capable (e.g. fine-grain data access, native object API, key-value store) computational storage than previous generation storage for HPC applications and workflows. <\/p>\n\n\n\n<p>Find <a rel=\"noreferrer noopener\" aria-label=\"more information (opens in a new tab)\" href=\"https:\/\/www.intel.com\/content\/www\/us\/en\/high-performance-computing\/daos-high-performance-storage-brief.html\" target=\"_blank\">more information<\/a> about how DAOS revolutionizes high-performance storage with Intel Optane DC Persistent Memory. <\/p>\n\n\n\n<p>[1]  <a href=\"https:\/\/www.marketscreener.com\/INTEL-CORPORATION-4829\/news\/Intel-Data-Centric-Portfolio-Accelerates-Convergence-of-High-Performance-Computing-and-AI-Workload-28774002\/\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\"https:\/\/www.marketscreener.com\/INTEL-CORPORATION-4829\/news\/Intel-Data-Centric-Portfolio-Accelerates-Convergence-of-High-Performance-Computing-and-AI-Workload-28774002\/ (opens in a new tab)\">https:\/\/www.marketscreener.com\/INTEL-CORPORATION-4829\/news\/Intel-Data-Centric-Portfolio-Accelerates-Convergence-of-High-Performance-Computing-and-AI-Workload-28774002\/<\/a><\/p>\n\n\n\n<p>[2]  <a href=\"https:\/\/lenovopress.com\/lp1083.pdf\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\"https:\/\/lenovopress.com\/lp1083.pdf (opens in a new tab)\">https:\/\/lenovopress.com\/lp1083.pdf<\/a> <\/p>\n\n\n\n<p>[3]  <a href=\"https:\/\/ucsdnews.ucsd.edu\/pressrelease\/intels_optane_dimms\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\"https:\/\/ucsdnews.ucsd.edu\/pressrelease\/intels_optane_dimms (opens in a new tab)\">https:\/\/ucsdnews.ucsd.edu\/pressrelease\/intels_optane_dimms<\/a><\/p>\n\n\n\n<p>[4]  <a href=\"https:\/\/arxiv.org\/pdf\/1903.05714.pdf\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\"https:\/\/arxiv.org\/pdf\/1903.05714.pdf (opens in a new tab)\">https:\/\/arxiv.org\/pdf\/1903.05714.pdf<\/a> <\/p>\n\n\n\n<p>[5]  <a href=\"https:\/\/www.research.ed.ac.uk\/portal\/en\/publications\/an-early-evaluation-of-intels-optane-dc-persistent-memory-module-and-its-impact-on-highperformance-scientific-applications(00807d1c-b95f-449a-8aaf-7e95d8fb6e45).html\" target=\"_blank\" rel=\"noreferrer noopener\" aria-label=\"https:\/\/www.research.ed.ac.uk\/portal\/en\/publications\/an-early-evaluation-of-intels-optane-dc-persistent-memory-module-and-its-impact-on-highperformance-scientific-applications(00807d1c-b95f-449a-8aaf-7e95d8fb6e45).html (opens in a new tab)\">https:\/\/www.research.ed.ac.uk\/portal\/en\/publications\/an-early-evaluation-of-intels-optane-dc-persistent-memory-module-and-its-impact-on-highperformance-scientific-applications(00807d1c-b95f-449a-8aaf-7e95d8fb6e45).html<\/a> <\/p>\n\n\n\n<p><a>For more complete information about performance and benchmark results, visit&nbsp;<\/a><strong><a rel=\"noreferrer noopener\" aria-label=\" (opens in a new tab)\" href=\"http:\/\/www.intel.com\/benchmarks\" target=\"_blank\">www.intel.com\/benchmarks<\/a><\/strong>.<\/p>\n\n\n\n<p>Intel technologies\u2019 features and benefits depend on system\nconfiguration and may require enabled hardware, software or service activation.\nPerformance varies depending on system configuration. Check with your system\nmanufacturer or retailer or learn more at intel.com.<\/p>\n\n\n\n<p>Intel, the Intel logo, and\nOptane are trademarks of Intel Corporation or its subsidiaries.<\/p>\n\n\n\n<p>Other names and brands\nmay be claimed as the property of others.&nbsp;<\/p>\n\n\n\n<p>\u00a9 Intel Corporation.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Forget what you previously knew about high-performance storage and file systems. New I\/O models for HPC such as Distributed Asynchronous Object Storage (DAOS) have been architected from the ground up to make use of new NVM technologies such as Intel\u00ae Optane\u2122 DC Persistent Memory Modules (Intel Optane DCPMMs). With latencies measured in nanoseconds and bandwidth measured in tens of GB\/s, new storage devices such as Intel DCPMMs redefine the measures used to describe high-performance nonvolatile storage.<\/p>\n","protected":false},"author":10513,"featured_media":22952,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"footnotes":""},"categories":[68,87,180,258,210,773,56,311,1],"tags":[429,284,788,774,364,95],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v20.6 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement - insideBIGDATA<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement - insideBIGDATA\" \/>\n<meta property=\"og:description\" content=\"Forget what you previously knew about high-performance storage and file systems. New I\/O models for HPC such as Distributed Asynchronous Object Storage (DAOS) have been architected from the ground up to make use of new NVM technologies such as Intel\u00ae Optane\u2122 DC Persistent Memory Modules (Intel Optane DCPMMs). With latencies measured in nanoseconds and bandwidth measured in tens of GB\/s, new storage devices such as Intel DCPMMs redefine the measures used to describe high-performance nonvolatile storage.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/\" \/>\n<meta property=\"og:site_name\" content=\"insideBIGDATA\" \/>\n<meta property=\"article:publisher\" content=\"http:\/\/www.facebook.com\/insidebigdata\" \/>\n<meta property=\"article:published_time\" content=\"2019-09-23T15:30:57+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2019-10-07T18:51:23+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/07\/shutterstock_1396230830.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"500\" \/>\n\t<meta property=\"og:image:height\" content=\"334\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Editorial Team\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@insideBigData\" \/>\n<meta name=\"twitter:site\" content=\"@insideBigData\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Editorial Team\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"6 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/\",\"url\":\"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/\",\"name\":\"DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement - insideBIGDATA\",\"isPartOf\":{\"@id\":\"https:\/\/insidebigdata.com\/#website\"},\"datePublished\":\"2019-09-23T15:30:57+00:00\",\"dateModified\":\"2019-10-07T18:51:23+00:00\",\"author\":{\"@id\":\"https:\/\/insidebigdata.com\/#\/schema\/person\/2949e412c144601cdbcc803bd234e1b9\"},\"breadcrumb\":{\"@id\":\"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/insidebigdata.com\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/insidebigdata.com\/#website\",\"url\":\"https:\/\/insidebigdata.com\/\",\"name\":\"insideBIGDATA\",\"description\":\"Your Source for AI, Data Science, Deep Learning &amp; Machine Learning Strategies\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/insidebigdata.com\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"en-US\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/insidebigdata.com\/#\/schema\/person\/2949e412c144601cdbcc803bd234e1b9\",\"name\":\"Editorial Team\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/insidebigdata.com\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/e137ce7ea40e38bd4d25bb7860cfe3e4?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/e137ce7ea40e38bd4d25bb7860cfe3e4?s=96&d=mm&r=g\",\"caption\":\"Editorial Team\"},\"sameAs\":[\"http:\/\/www.insidebigdata.com\"],\"url\":\"https:\/\/insidebigdata.com\/author\/editorial\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement - insideBIGDATA","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/","og_locale":"en_US","og_type":"article","og_title":"DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement - insideBIGDATA","og_description":"Forget what you previously knew about high-performance storage and file systems. New I\/O models for HPC such as Distributed Asynchronous Object Storage (DAOS) have been architected from the ground up to make use of new NVM technologies such as Intel\u00ae Optane\u2122 DC Persistent Memory Modules (Intel Optane DCPMMs). With latencies measured in nanoseconds and bandwidth measured in tens of GB\/s, new storage devices such as Intel DCPMMs redefine the measures used to describe high-performance nonvolatile storage.","og_url":"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/","og_site_name":"insideBIGDATA","article_publisher":"http:\/\/www.facebook.com\/insidebigdata","article_published_time":"2019-09-23T15:30:57+00:00","article_modified_time":"2019-10-07T18:51:23+00:00","og_image":[{"width":500,"height":334,"url":"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/07\/shutterstock_1396230830.jpg","type":"image\/jpeg"}],"author":"Editorial Team","twitter_card":"summary_large_image","twitter_creator":"@insideBigData","twitter_site":"@insideBigData","twitter_misc":{"Written by":"Editorial Team","Est. reading time":"6 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/","url":"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/","name":"DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement - insideBIGDATA","isPartOf":{"@id":"https:\/\/insidebigdata.com\/#website"},"datePublished":"2019-09-23T15:30:57+00:00","dateModified":"2019-10-07T18:51:23+00:00","author":{"@id":"https:\/\/insidebigdata.com\/#\/schema\/person\/2949e412c144601cdbcc803bd234e1b9"},"breadcrumb":{"@id":"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/insidebigdata.com\/2019\/09\/23\/daos-delivers-exascale-performance-using-hpc-storage-so-fast-it-requires-new-units-of-measurement\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/insidebigdata.com\/"},{"@type":"ListItem","position":2,"name":"DAOS Delivers Exascale Performance Using HPC Storage So Fast It Requires New Units of Measurement"}]},{"@type":"WebSite","@id":"https:\/\/insidebigdata.com\/#website","url":"https:\/\/insidebigdata.com\/","name":"insideBIGDATA","description":"Your Source for AI, Data Science, Deep Learning &amp; Machine Learning Strategies","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/insidebigdata.com\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/insidebigdata.com\/#\/schema\/person\/2949e412c144601cdbcc803bd234e1b9","name":"Editorial Team","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/insidebigdata.com\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/e137ce7ea40e38bd4d25bb7860cfe3e4?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/e137ce7ea40e38bd4d25bb7860cfe3e4?s=96&d=mm&r=g","caption":"Editorial Team"},"sameAs":["http:\/\/www.insidebigdata.com"],"url":"https:\/\/insidebigdata.com\/author\/editorial\/"}]}},"jetpack_featured_media_url":"https:\/\/insidebigdata.com\/wp-content\/uploads\/2019\/07\/shutterstock_1396230830.jpg","jetpack_shortlink":"https:\/\/wp.me\/p9eA3j-640","jetpack-related-posts":[{"id":23428,"url":"https:\/\/insidebigdata.com\/2019\/10\/17\/six-platform-investments-from-intel-to-facilitate-running-ai-and-hpc-workloads-together-on-existing-infrastructure\/","url_meta":{"origin":23312,"position":0},"title":"Six Platform Investments from Intel to Facilitate Running AI and HPC Workloads Together on Existing Infrastructure","date":"October 17, 2019","format":false,"excerpt":"Because HPC technologies today offer substantially more power and speed than their legacy predecessors, enterprises and research institutions benefit from combining AI and HPC workloads on a single system. Six platform investments from Intel will help reduce obstacles and make HPC and AI deployment even more accessible and practical.","rel":"","context":"In &quot;AI Deep Learning&quot;","img":{"alt_text":"","src":"https:\/\/i0.wp.com\/insidebigdata.com\/wp-content\/uploads\/2019\/07\/shutterstock_1396230830.jpg?resize=350%2C200&ssl=1","width":350,"height":200},"classes":[]},{"id":23579,"url":"https:\/\/insidebigdata.com\/2019\/11\/18\/2nd-generation-intel-xeon-platinum-9200-processors-offer-leadership-performance-and-advance-ai\/","url_meta":{"origin":23312,"position":1},"title":"2nd Generation Intel\u00ae Xeon\u00ae Platinum 9200 Processors Offer Leadership Performance, and Advance AI","date":"November 18, 2019","format":false,"excerpt":"Simulation, modeling, data analytics, and other workloads commonly use high performance computing (HPC) to advance research and business in many ways. However, as converged workloads involving AI grow in adoption, HPC systems must keep pace with evolving needs. 2nd Generation Intel\u00ae Xeon\u00ae Platinum processors, with built-in AI acceleration technologies, offer\u2026","rel":"","context":"In &quot;AI Deep Learning&quot;","img":{"alt_text":"","src":"https:\/\/i0.wp.com\/insidebigdata.com\/wp-content\/uploads\/2019\/11\/xeon-pltnm-m-ww-rgb-3000.png?resize=350%2C200&ssl=1","width":350,"height":200},"classes":[]},{"id":10763,"url":"https:\/\/insidebigdata.com\/2014\/08\/13\/intel-enterprise-edition-software-lustre\/","url_meta":{"origin":23312,"position":2},"title":"Intel Enterprise Edition Software for Lustre","date":"August 13, 2014","format":false,"excerpt":"Intel Enterprise Edition Software for Lustre can transform vast amounts of data into data-driven decisions. And using the software with Hadoop makes storage management simpler with single Lustre file systems rather than partitioned, hard-to-manage storage.","rel":"","context":"In &quot;Big Data Software&quot;","img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]},{"id":10757,"url":"https:\/\/insidebigdata.com\/2014\/08\/08\/intel-cloud-edition-available-lustre-software\/","url_meta":{"origin":23312,"position":3},"title":"Sponsored Post: Intel Cloud Edition Available for Lustre Software","date":"August 8, 2014","format":false,"excerpt":"Intel has collaborated with AWS to offer a Cloud Edition for Lustre Software that allows customers to use the power of the worlds' most popular HPC storage system. It provides fast, scalable storage, optimizing servers for the workload they support.","rel":"","context":"In &quot;Big Data&quot;","img":{"alt_text":"","src":"https:\/\/i0.wp.com\/insidebigdata.com\/wp-content\/uploads\/2014\/08\/Lustreonaws2.png?resize=350%2C200&ssl=1","width":350,"height":200},"classes":[]},{"id":17008,"url":"https:\/\/insidebigdata.com\/2017\/01\/26\/hpc-storage-performance-in-the-cloud\/","url_meta":{"origin":23312,"position":4},"title":"HPC Storage Performance in the Cloud","date":"January 26, 2017","format":false,"excerpt":"In this contributed article, technical story teller Ken Strandberg, discusses the feeding of high-performance computing (HPC) and enterprise technical computing clusters with data using Lustre, the open source parallel file system that provides the performance and scalability to meet the demands of workloads on these systems.","rel":"","context":"In &quot;Cloud&quot;","img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]},{"id":13691,"url":"https:\/\/insidebigdata.com\/2015\/09\/15\/enabling-value-for-converged-commercial-hpc-and-big-data-infrastructures-through-lustre\/","url_meta":{"origin":23312,"position":5},"title":"Enabling Value for Converged Commercial HPC and Big Data Infrastructures through Lustre*","date":"September 15, 2015","format":false,"excerpt":"A number of industries rely on high-performance computing (HPC) clusters to process massive amounts of data. As these same organizations explore the value of Big Data analytics based on Hadoop, they are realizing the value of converging Hadoop and HPC onto the same cluster rather than scaling out an entirely\u2026","rel":"","context":"In &quot;Big Data Hardware&quot;","img":{"alt_text":"","src":"https:\/\/i0.wp.com\/insidebigdata.com\/wp-content\/uploads\/2015\/09\/big-data-infrastructure.png?resize=350%2C200&ssl=1","width":350,"height":200},"classes":[]}],"_links":{"self":[{"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/posts\/23312"}],"collection":[{"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/users\/10513"}],"replies":[{"embeddable":true,"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/comments?post=23312"}],"version-history":[{"count":0,"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/posts\/23312\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/media\/22952"}],"wp:attachment":[{"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/media?parent=23312"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/categories?post=23312"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/insidebigdata.com\/wp-json\/wp\/v2\/tags?post=23312"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}