HPC End Users Cite Mixed I/O as the Most Difficult Performance
Challenge and Burst Buffers Are Seen as the Next Step for High
Performance Data

SANTA CLARA, CA – Nov. 2, 2016 — /BackupReview.info/ — DataDirect Networks (DDN®) today announced the results of its annual High Performance Computing
(HPC) Trends survey, which showed that end users in the world’s most data-intense
environments, like those in many general IT environments, are increasing their
use of cloud. However, unlike general IT environments, the HPC sector is
overwhelmingly opting for private and hybrid clouds instead of public. They
are also increasingly choosing to upgrade specific parts of their environments
with flash as they modernize their data centers. Managing mixed I/O performance
and rapid data growth remain the biggest challenges for HPC organizations driving
these infrastructure changes according to the survey.

Conducted by DDN for the fourth consecutive year, the survey polled a cross-section
of 143 end-users managing data intensive infrastructures worldwide and representing
hundreds of petabytes of storage investment. Respondents included individuals
responsible for high performance computing and also networking and storage systems
from financial services, government, higher education, life sciences, manufacturing,
national labs, and oil and gas organizations. The volume of data under management
in each of these organizations is staggering and steadily increasing each year.

Of organizations surveyed:

  • 73 percent manage or use more than one petabyte of data storage; and
  • 30 percent manage or use more than 10 PBs of data storage,
    up 5 percentage points year-over-year.

 

 

2016 HPC Trends Survey Storage Use

 

Data and data storage remain the most strategic part of the HPC data center,
according to an overwhelming majority of survey respondents (77 percent), as end
users seek to solve data access, workflow and analytics challenges to accelerate time
to results.

Tweet This: Mixed I/O remains No. 1 #BigData challenge; Rising use of
#flash & hybrid/private #cloud in @DDN_Limitless #HPC survey http://bit.ly/2dLoiJ9

Survey respondents revealed the rising use of private and hybrid clouds within
HPC data centers. Respondents planning to leverage a cloud for at least part of their
data in 2017 rose to 37 percent, up almost 10 percentage points year-over-year.
Of those, more than 80 percent are choosing private or hybrid clouds versus a public
cloud option. “These responses are consistent with the trends DDN observes in
customer conversations, the most prevalent of which is organizations rebounding from
public cloud due to cost, poor latency and sheer data immobility issues,” said
Laura Shepard, senior director of marketing, DDN.

Use of flash in HPC data centers has intensified with more than 90 percent of respondents using flash storage at some level within their data centers today. Perhaps surprisingly, while all-flash arrays are perceived by many to be the fastest storage available in the market, only 10 percent of surveyed users from these most data-intense environments are using an all-flash array. The vast majority of respondents (80 percent) are using hybrid flash arrays either as an extension to storage-level cache, to accelerate metadata, or to accelerate data sets associated with key or problem I/O applications.

2016 HPC Trends Survey Current Flash Storage Use

A diverse set of applications and an upsurge in site-wide file systems, data lakes and active archives are driving fast-paced data growth in large-scale environments and analytical workflows, which are placing rigorous demands on storage infrastructures and creating unique challenges for HPC users. Performance ranks as the number one storage and Big Data challenge by strong majority (76 percent) of those polled; and mixed I/O performance was cited as the biggest concern by a strong majority (61 percent) of the respondents, which represented an eight percentage point increase compared with last year’s survey results.

2016 HPC Trends Survey Big Data Challenge

An even higher portion of respondents (68 percent) identify storage I/O as the main bottleneck specifically in analytics workflows. As these responses demonstrate, performance issues have escalated as Big Data environments contend with a proliferation of diverse applications creating mixed I/O patterns and stifling the performance of their storage infrastructure.Only a small and diminishing percentage of respondents believe today’s file systems and data management technologies will be able to scale to Exascale levels, while almost 75 percent of respondents believe new innovation will be required. This belief is illustrated in respondents’ views on addressing performance issues:

  • The strong majority (60 percent) view burst buffers to be the most likely technology to take storage to the next level as users seek faster and more efficient ways to offload I/O from compute resources, to separate storage bandwidth acquisition from capacity acquisition and to support parallel file systems to meet Exascale requirements.

As an increasing number of HPC sites move to real-world implementation of multi-site HPC collaboration, concerns about security, privacy, and data sharing have intensified significantly. A strong majority of those surveyed (70 percent) view security and data-sharing complexity as the biggest impediments to multi-site collaborations.

  • One HPC organization that is at the forefront of employing leading-edge technologies to keep its research data secure and private is Weill Cornell Medicine, a large, multi-site organization managing massive volumes of patient data. To watch a video of Vanessa Borcherding, Weill Cornell Medicine’s Director of Scientific Computing Unit, discuss their approach to securing potentially sensitive genomic data that needs to be available for collaborative research, click here.

With storage performance a critical requirement for today’s large-scale and petascale-level data centers, site-wide file systems continue to be a significant infrastructure trend in HPC environments, according to 72 percent of HPC customers polled. Site-wide file systems allow architects either to consolidate data storage multiple compute clusters on the same storage platform and/or to provide the flexibility to upgrade storage and servers independently as needed. In addition to some of the largest supercomputing sites like those at Oakridge National Lab (ORNL), National Energy Research Scientific Computing Center (NERSC) and Texas Advanced Computing Center (TACC), site-wide file systems are expanding into more mid-sized data centers with a smaller number or smaller sized compute clusters.

“The results of DDN’s annual HPC Trends Survey reflect very accurately what HPC end users tell us and what we are seeing in their data center infrastructures. The use of private and hybrid clouds continues to grow although most HPC organizations are not storing as large a percentage of their data in public clouds as they anticipated even a year ago. Performance remains the top challenge, especially when handling mixed I/O workloads and resolving I/O bottlenecks. Given this, it’s not surprising that 90 percent of those surveyed are using flash within their data centers, but what is notable is that the more storage experience a site has, the more likely they are to use flash to accelerate multiple tiers of storage rather than putting it all in one tier for one part of the workflow,” added Shepard. “Survey respondents also reaffirmed that data storage is the most strategic part of the data center.”

Upcoming SC16 Conference
DDN executives, senior level engineering staff and industry workflow experts will be available at the SC16 conference in Salt Lake City, Utah, at DDN booth #1931 to discuss these HPC trends and the latest DDN products and innovations. To schedule a meeting, click here.

Supporting Resources

  • Survey results available upon request at pr@ddn.com
  • More on DDN at SC16
  • More on DDN HPC Storage Solutions
  • More on DDN Customers

Follow DDN: Blog, Twitter, LinkedIn and Facebook

About DDN
DataDirect Networks (DDN) is the world’s leading big data storage supplier to data-intensive, global organizations. For more than 15 years, DDN has designed, developed, deployed and optimized systems, software and storage solutions that enable enterprises, service providers, universities and government agencies to generate more value and to accelerate time to insight from their data and information, on premise and in the cloud. Organizations leverage the power of DDN storage technology and the deep technical expertise of its team to capture, store, process, analyze, collaborate and distribute data, information and content at largest scale in the most efficient, reliable and cost-effective manner. DDN customers include many of the world’s leading financial services firms and banks, healthcare and life science organizations, manufacturing and energy companies, government and research facilities, and web and cloud service providers. For more information, go to ddn.com or call 1-800-837-2298.

Media Contacts:
DDN
Michael King
Sr. Director of Marketing
mking@ddn.com
408-419-2892

IGNITE Consulting, on behalf of DDN
Linda Dellett, 303-439-9398
Kathleen Sullivan, 303-439-9365
ddn@igniteconsultinginc.com

©2016 All rights reserved. DDN Storage and DDN are trademarks owned by DataDirect Networks. All other trademarks are the property of their respective owners.

Source: DDN

Tags: