HPE contributes to improve the DAOS experience for all users

[AdvEXperts]

Learn how a collaboration between Intel and HPE is improving the usability and experience of the open-source DAOS software stack and what admins and users can expect from the new release.

By Kelsey Prantis, Senior Software Engineering Manager at Intel

Contributions from a multiyear collaboration between Intel and HPE that enhance both usability and the customer experience will be added to the open source Distributed Asynchronous Object Storage (DAOS) 2.0 release expected to be distributed before the end of the year. As an active contributor to the DAOS software project, we are grateful to our industry partners and expect that HPE’s participation and extensive experience will continue to refine future releases of the open-source DAOS software stack to the benefit of all users. Such contributions reflect the value of the many DAOS contributors who are driving the rapid rate of adoption and maturation of the DAOS software.

DAOS is the open source software scale-out object store that leverages the latest Intel technology to provide high bandwidth, low latency, and high I/O operations per second (IOPS) storage containers. Data intensive computing has made storage both a performance focal point and critical bottleneck in data centers around the world. DAOS rethinks the traditional operating system and file system approach by giving applications direct access to storage from user-space. That, coupled with byte-addressable nature of the Intel® Optane™ media which can be package as memory based Intel persistent memory devices, or the NVMe-based Intel Optane storage devices means that users can now benefit from  several orders of magnitude faster access to storage and world record setting storage performance. [1] DAOS also makes use of the latest RDMA capable fabrics to maximize performance and scalability in distribute computing environments. [2]

Part of a broader collaboration, HPE has worked with Intel to address the needs of both DAOS administrators and users. HPE’s global reach and extensive across-the-board experience with data centers of all types and sizes has facilitated a critical and immensely practical review of DAOS functionality. Even the syntax of the DAOS commands have undergone intense scrutiny. The end result is that DAOS 2.0 provides a more consistent naming scheme for the command structure, which in turn, makes it easier for people to use and script DAOS to fit their particular needs.

Further, all DAOS users will appreciate the new client-side metrics that HPE contributed as open source to the DAOS software project. Based on feedback from HPE, Intel has made a number of other smaller—but still important—modifications such as the switch to human-understandable labels rather than the 128-bit UUID numbers. Cumulatively, these changes provide a more intuitive interface and make it easier for both users and systems managers to evaluate and tune how they use DAOS and the Intel Optane technology.

My recent video on the DAOS Ecosystem highlights how DAOS overcomes a number of industry bottlenecks that inhibited users accessing the full capability of the Intel Optane Persistent Memory modules. The recently announced 200 series of devices are both performant and extremely flexible as they can plug directly into the memory bus of select Intel Xeon® Scalable processors to augment main system memory, be used as DAOS storage, or partitioned to support both at the same time. [3]

Looking to the future, upcoming versions of DAOS will include improvements that will open up new possibilities to differentiate performance across a range of NVMe media classes that are available now such as the near-nanosecond latency Intel Optane SSD devices and NAND-based SSDs. 

With DAOS, data centers can realize the performance of the latest in Intel technology while also mixing and matching their Intel Optane devices to best serve their customers storage and memory capacity needs.

Learn more about DAOS

Access the DAOS open source

Visit the Discover More Network to hear the latest advancement in enabling HPC and AI at HPE 

---

Meet our guest blogger Kelsey Prantiss, Senior Software Engineering Manager at Intel

Kelsey Prantis heads the Extreme Storage Architecture and Development team inside of the Accelerated Computing Systems and Graphics (AXG) Group. She leads the development of Distributed Asynchronous Object Storage (DAOS), an open-source, low-latency and high IOPS scale-out object store designed from the ground up for massively distributed Non-Volatile Memory (NVM), which is revolutionizing high-performance storage for HPC, AI, and Big Data, both on prem and in the cloud. She joined Intel in 2012 with the acquisition of Whamcloud, where she led the development of the Intel Manager for Lustre* product. Prior to Whamcloud, she was a software developer at personal genomics and biotechnology company 23andMe. Prantis holds a Bachelor’s degree in Computer Science from Rochester Institute of Technology.

Advantage EX Experts
Hewlett Packard Enterprise

twitter.com/hpe_hpc
linkedin.com/showcase/hpe-ai/
hpe.com/info/hpc

Intel technologies may require enabled hardware, software or service activation. 

No product or component can be absolutely secure.

© Intel Corporation.  Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.  Other names and brands may be claimed as the property of others.  

[1] https://www.intel.com/content/www/us/en/developer/articles/technical/daos-sets-new-records-with-intel-optane-persistent-memory.html
[2] https://itpeernetwork.intel.com/fabrics-transport-protocols/
[3] https://www.intel.com/content/dam/support/us/en/documents/memory-and-storage/data-center-persistent-mem/Intel_Optane_Persistent_Memory_Start_Up_Guide.pdf