0
0
1,735
On Thursday, Aug 6, the Intel Optane team was given the opportunity to share our vision of the evolving memory and storage landscape at Storage Field Day 20. The industry analysts and influencers in attendance gave us a very warm welcome and asked a lot of great questions as well as provided their insights throughout the session. After years of development, planning, and engineering, the potential that Intel Optane technology has always promised is now manifested in a series of new products and solutions, performance benchmarks, and a supporting ecosystem of partner technologies. As we’ve learned, when you launch a new memory and storage architecture it takes time to get all the pieces in place and all the partner ecosystems aligned, but it was clear from our reception at Storage Field Day that the vision of a new data-centric approach to memory and storage architectures is becoming a reality.
Storage Field Day is part of the Tech Field Day series of invite-only technical meetings that bring together journalists and analysts from around the world, and executives from enterprise IT companies to discuss new products and ideas through: presentations, demos, roundtables, and panel discussions. The Field Day events are live-streamed and recorded, watch our Storage Field Day event here.
Our session kicked off with a presentation by Chris Tobias, the General Manager of the Intel Optane Software Solutions Group. He provided an under-the-hood look into the construction and characteristics of the Intel Optane media technology. He described how Intel Optane is a unique media combining the best characteristics of DRAM (byte addressability) and NAND (persistence). This enables a new tier in the memory/storage data hierarchy. Watch Chris’s primer on Intel Optane technology:
Chris went on to share how the Intel Optane SSD team is working with customers to provide real value today. Intel Optane SSDs have proven to be especially valuable in deployments for software-defined storage (SDS) and hyper-converged infrastructure (HCI).
For instance, Optane SSDs are an excellent metadata drive for Ceph, the popular open-source SDS file system. Adding a relatively small amount of Optane storage to the Ceph BlueStore file system can greatly reduce average and tail latencies. More and more clients realize Intel’s open-source block storage caching software is a great way to accelerate a Ceph or any Storage environment where I/O wait times are a major hurdle. NAND-based SSDs simply cannot keep up with the speed and core counts of today’s high-performance CPUs.
Chris also described how our team worked with the Cisco Hyperflex team to create an HCI solution that has delivered some amazing results for the Los Angeles Dodgers. In fact, the results for the Dodgers speak for themselves. Analytics processing was accelerated by 13x, and IT footprint and licensing was reduced by over 70 percent. The IT team was able to shift its focus from maintenance to growth and innovation, and analytics processing that used to take 13 hours is now completed in less than one hour.
Chris wrapped up the discussion on HCI with a discussion on how Intel Optane SSDs improve VMware vSAN usages. Using an Intel Optane SSD can enable a 33 percent node reduction, 60 percent increase in performance in database applications, and a 15 percent improvement in day-to-day operations like cloning, snapshots and storage vMotion. Cerner, one of the largest healthcare IT providers in the US, was extremely pleased with the remarkable responsiveness provided by adding Intel Optane SSDs to vSAN. Watch Chris’s presentation on Intel Optane SSDs:
Next up, I shared a deeper dive on our persistent memory products and provided an example of HPC storage architecture innovations based on Optane persistent memory.
We started with an overview of the product and the latest features of our 2nd generation product, which we’re calling the 200 Series, launched in June. I explained that we have implemented a number of controller and power/performance improvements in the new release that allow us to provide better performance at even lower power levels than the previous generation of persistent memory modules, increasing the performance/TCO value to our customers. With these improvements, and with elements of Intel Optane media, the DDR-T protocol, persistent memory firmware, operating modes, memory controllers, and CPU platform integration, our Intel Optane persistent memory device is a full solution on a DIMM. The analysts even commented that Intel had done a good job of including the software ecosystem and platform integration elements.
But the real conversations got started when we shared an innovative new use case for HPC and analytics based on persistent memory and a new open-source software stack called DAOS (Distributed Asynchronous Object Store).
We told the story of how Intel built a new distributed storage solution using the DAOS software stack and Intel Optane persistent memory. The high point of this story was that Texas Advanced Computing Center, Argonne National Laboratory, and Intel, each built prototype architectures on this distributed storage solution and took the top three spots on the IO500 10 Node Challenge list for File System performance. This performance is made possible due to the unique hardware capabilities of Intel Optane persistent memory: combining low-latency byte-granular memory access with data persistence for small I/Os, and the new DAOS software stack built from the ground up to leverage these capabilities. Watch my presentation on Intel Optane persistent memory usages:
The analysts were impressed by this world-record-setting performance, and Kelsey Prantis, DAOS senior software engineering manager, joined us to provide a dive deep into how we did it. While Intel was excited about these results, it’s fair to say the event participants were blown away. They provided a lot of insights to the Intel team around how this level of innovation and performance can shape additional new usages. One of the analysts said, and I paraphrase, “You’re innovating and providing a showcase of what other people can build on this foundation. And that’s really important when it comes to hardware technology. If you show people how to make the most of the new hardware, then the rest of the industry will follow and flourish.”
Next, Frank Ober, principal engineer for Intel Optane technologies at Intel, presented a fascinating case study on how high performance Intel Optane persistent memory is being used in the cloud. VK is Russia’s largest social media network, with over 500 million user accounts and roughly 97 million active users a month. VK turned to Intel and Optane technology to provide a refresh of their infrastructure and help them bring down compute and infrastructure costs. They implemented a two-tiered web and content delivery network cache system using Optane SSD’s and Intel QLC 3D NAND SSD’s for storage with Intel FPGA’s as video transcoders and Optane PMem for real-time data processing in memory using Memcached (an open-source caching software).
The result of their infrastructure refresh was a densification of the VK platform, along with a 40 percent reduction of server count in their data center and higher performance across their entire infrastructure. In addition, because many elements of the Intel portfolio of offerings were included in the solution, it was easy for them to adopt. It just worked! Watch Frank’s presentation on Intel Optane technology in the cloud:
We’ve been on this journey with Intel Optane technology and driving new capabilities like persistent memory for such a long time, and it’s exciting to see our vision becoming a reality. The ecosystem is really coming together: most of the major storage solutions are optimized for Optane and all the major operating systems are now persistent memory-aware, as are most server products and database infrastructure applications. And we’re starting to see more and more new applications, like DAOS, emerging each day, built on persistent memory’s potential to radically realign the building blocks of traditional computing.
When you think about it, in this industry we’ve accepted the large gap between memory and storage, performance and costs, and have built all sorts of workarounds and trade-offs to deal with it. But now we’re asking, what if you didn’t have that gap, and that latency? What opportunities does that open up? What could you do with that? With Intel Optane technology, the horizon is open, so go get creative!
Follow @Kristie_L_Mann on Twitter for the latest news and views on disruptive technology, persistent memory and evolving data center architectures.
Storage Field Day is part of the Tech Field Day series of invite-only technical meetings that bring together journalists and analysts from around the world, and executives from enterprise IT companies to discuss new products and ideas through: presentations, demos, roundtables, and panel discussions. The Field Day events are live-streamed and recorded, watch our Storage Field Day event here.
Building an understanding of the Optane Technology
Our session kicked off with a presentation by Chris Tobias, the General Manager of the Intel Optane Software Solutions Group. He provided an under-the-hood look into the construction and characteristics of the Intel Optane media technology. He described how Intel Optane is a unique media combining the best characteristics of DRAM (byte addressability) and NAND (persistence). This enables a new tier in the memory/storage data hierarchy. Watch Chris’s primer on Intel Optane technology:
Creating business value with proven Intel Optane SSDs
Chris went on to share how the Intel Optane SSD team is working with customers to provide real value today. Intel Optane SSDs have proven to be especially valuable in deployments for software-defined storage (SDS) and hyper-converged infrastructure (HCI).
For instance, Optane SSDs are an excellent metadata drive for Ceph, the popular open-source SDS file system. Adding a relatively small amount of Optane storage to the Ceph BlueStore file system can greatly reduce average and tail latencies. More and more clients realize Intel’s open-source block storage caching software is a great way to accelerate a Ceph or any Storage environment where I/O wait times are a major hurdle. NAND-based SSDs simply cannot keep up with the speed and core counts of today’s high-performance CPUs.
Chris also described how our team worked with the Cisco Hyperflex team to create an HCI solution that has delivered some amazing results for the Los Angeles Dodgers. In fact, the results for the Dodgers speak for themselves. Analytics processing was accelerated by 13x, and IT footprint and licensing was reduced by over 70 percent. The IT team was able to shift its focus from maintenance to growth and innovation, and analytics processing that used to take 13 hours is now completed in less than one hour.
Chris wrapped up the discussion on HCI with a discussion on how Intel Optane SSDs improve VMware vSAN usages. Using an Intel Optane SSD can enable a 33 percent node reduction, 60 percent increase in performance in database applications, and a 15 percent improvement in day-to-day operations like cloning, snapshots and storage vMotion. Cerner, one of the largest healthcare IT providers in the US, was extremely pleased with the remarkable responsiveness provided by adding Intel Optane SSDs to vSAN. Watch Chris’s presentation on Intel Optane SSDs:
Catching up on Intel Optane persistent memory news
Next up, I shared a deeper dive on our persistent memory products and provided an example of HPC storage architecture innovations based on Optane persistent memory.
We started with an overview of the product and the latest features of our 2nd generation product, which we’re calling the 200 Series, launched in June. I explained that we have implemented a number of controller and power/performance improvements in the new release that allow us to provide better performance at even lower power levels than the previous generation of persistent memory modules, increasing the performance/TCO value to our customers. With these improvements, and with elements of Intel Optane media, the DDR-T protocol, persistent memory firmware, operating modes, memory controllers, and CPU platform integration, our Intel Optane persistent memory device is a full solution on a DIMM. The analysts even commented that Intel had done a good job of including the software ecosystem and platform integration elements.
But the real conversations got started when we shared an innovative new use case for HPC and analytics based on persistent memory and a new open-source software stack called DAOS (Distributed Asynchronous Object Store).
We told the story of how Intel built a new distributed storage solution using the DAOS software stack and Intel Optane persistent memory. The high point of this story was that Texas Advanced Computing Center, Argonne National Laboratory, and Intel, each built prototype architectures on this distributed storage solution and took the top three spots on the IO500 10 Node Challenge list for File System performance. This performance is made possible due to the unique hardware capabilities of Intel Optane persistent memory: combining low-latency byte-granular memory access with data persistence for small I/Os, and the new DAOS software stack built from the ground up to leverage these capabilities. Watch my presentation on Intel Optane persistent memory usages:
The analysts were impressed by this world-record-setting performance, and Kelsey Prantis, DAOS senior software engineering manager, joined us to provide a dive deep into how we did it. While Intel was excited about these results, it’s fair to say the event participants were blown away. They provided a lot of insights to the Intel team around how this level of innovation and performance can shape additional new usages. One of the analysts said, and I paraphrase, “You’re innovating and providing a showcase of what other people can build on this foundation. And that’s really important when it comes to hardware technology. If you show people how to make the most of the new hardware, then the rest of the industry will follow and flourish.”
Intel Optane SSDs and persistent memory provide high performance in the cloud
Next, Frank Ober, principal engineer for Intel Optane technologies at Intel, presented a fascinating case study on how high performance Intel Optane persistent memory is being used in the cloud. VK is Russia’s largest social media network, with over 500 million user accounts and roughly 97 million active users a month. VK turned to Intel and Optane technology to provide a refresh of their infrastructure and help them bring down compute and infrastructure costs. They implemented a two-tiered web and content delivery network cache system using Optane SSD’s and Intel QLC 3D NAND SSD’s for storage with Intel FPGA’s as video transcoders and Optane PMem for real-time data processing in memory using Memcached (an open-source caching software).
The result of their infrastructure refresh was a densification of the VK platform, along with a 40 percent reduction of server count in their data center and higher performance across their entire infrastructure. In addition, because many elements of the Intel portfolio of offerings were included in the solution, it was easy for them to adopt. It just worked! Watch Frank’s presentation on Intel Optane technology in the cloud:
Opening the door to innovation
We’ve been on this journey with Intel Optane technology and driving new capabilities like persistent memory for such a long time, and it’s exciting to see our vision becoming a reality. The ecosystem is really coming together: most of the major storage solutions are optimized for Optane and all the major operating systems are now persistent memory-aware, as are most server products and database infrastructure applications. And we’re starting to see more and more new applications, like DAOS, emerging each day, built on persistent memory’s potential to radically realign the building blocks of traditional computing.
When you think about it, in this industry we’ve accepted the large gap between memory and storage, performance and costs, and have built all sorts of workarounds and trade-offs to deal with it. But now we’re asking, what if you didn’t have that gap, and that latency? What opportunities does that open up? What could you do with that? With Intel Optane technology, the horizon is open, so go get creative!
Follow @Kristie_L_Mann on Twitter for the latest news and views on disruptive technology, persistent memory and evolving data center architectures.
About the Author
Kristie Mann is VP/GM for the IPU Business Unit in Intel’s Network and Edge Group. She focuses on developing the ecosystem for networking infrastructure and storage offload products. She was previously VP Product Management for Optane Products in Intel’s Datacenter and AI Group. Before joining Intel, she was Sr Manager R&D Hardware at HPE, where she was responsible for delivering mechanical hardware and thermal systems development for workstation lines that produced $1.2B of revenue annually. She earned a BSME at Georgia Tech and an MBA at Duke University. She has been a presenter at Tech Field Day, Global Stac Live and SmartNICs Summit and has been interviewed or quoted in many leading technology media such as Blocks & Files, Futurum Research and InfoWorld.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.