How Intel Is Making Optane Memory Easier For Partners To Sell

‘We‘re going to help with providing the solutions, the training or even offering programs or incentives for them to become champions, so that they become educated on the product and how to use it. And we’re really starting to see a lot of traction,’ Intel’s Kristie Mann says of Optane Persistent Memory for data centers.

Intel knows it can’t sell Intel Optane Persistent Memory like it does its processors, so the chipmaker is taking a solutions-based approach for the nascent technology and expanding its reach with partners to tout the performance and economics benefits for data centers.

The new memory tier, which combines the persistent qualities of storage with performance that nearly rivals DRAM, wasn’t available to channel partners except through OEM servers when the first generation launched in 2019 alongside Intel’s Cascade Lake server processors.

[Related: The $35B AMD-Xilinx Acquisition: 7 Big Things To Know]

But that has changed recently, according to Kristie Mann, senior director of Optane Persistent Memory products at Intel, who said the chipmaker worked with a handful of partners on proofs of concept before deciding to open the product line up to the broader channel several months ago.

‘We‘re going to help with providing the solutions, the training or even offering programs or incentives for them to become ‘champions,’ so that they become educated on the product and how to use it. And we’re really starting to see a lot of traction,” she said in a recent interview with CRN.

Optane Persistent Memory isn’t to be confused with Intel’s NAND memory business, which consists of NAND SSD products and is in the process of being sold to SK Hynix for $9 billion. Intel affirmed its support for Optane memory technologies last week when it announced the deal.

Mann said Optane Persistent Memory, which is onto its second generation that launched in June, is supported by more than 130 server platforms now, and that more than 200 of the Fortune 500 companies are running proofs of concepts or deployed the technology. Out of the proofs of concept, some 85 percent have converted to production environments.

“That‘s not normal for a new technology,” she said.

The product line has a “vibrant and growing ecosystem” of independent software vendors and operating software vendors, which includes SAP, Oracle and VMware. Mann said Intel has also been “working on upstreaming open-source software.” Then there’s a new wave of ISVs that are creating new software platforms based on Optane Persistent Memory, like MemVerge.

Mann said Intel is taking a solutions-based approach with selling Optane Persistent Memory through the channel because the product it isn’t as straightforward as a traditional Intel processor.

“What I like to say is no customer ever had a persistent memory problem, but they have business problems, and they want a solution to that problem,” she said.

“So we‘re developing key solutions with reference architectures and documentation on how to do it, guides on how to have the conversation,” Mann added. “What we’re seeing is that our partners are becoming that trusted adviser to go touch those customers, support them globally and really help them solve whatever problem it is that they’re trying to solve.”

These solutions can be easily found on the Optane Persistent Memory webpage, and they include solutions for things like infrastructure and virtualization, databases and analytics.

One classic problem Optane Persistent Memory can address is the long amount of time it can take to restart a server after power failure or when a patch needs to be applied. When a server reboots, the data leaves memory and then needs to be loaded again, which Mann said can take hours. But with the persistent quality of Intel Persistent Memory, that problem is no longer there.

This is made possible by Optane Persistent Memory’s App Direct Mode, which needs to be enabled by developers, whether in code or through an abstraction layer like MemVerge’s Memory Machine. Otherwise, Optane Persistent Memory will run in memory mode, which translates into “large capacities of affordable memory,” according to Mann.

“When I really talk about getting storage at memory speeds, that‘s what we’re talking about here,” she said. “You used to have to write it to a hard drive or a solid-state drive before you knew that [Optane Persistent Memory] was persistent and permanent. And all of those activities of duplicating the data because you have to have three copies and you always have to do that through the NVMe bus — which took a lot of time and it slowed things down — now we can bring that capability to do all of your storage functionality but in the memory subsystem.”

Charles Fan, CEO and co-founder of MemVerge, which counts Intel Capital as an investor, said his company’s Memory Machine platform can create giant pools of memory, whether it’s DRAM or Optane Persistent Memory, to provide a “virtualized, software-defined memory service” that gives applications larger memory capacities at lower costs.

“It can provide more memory for cheaper. This has the effect of lowering the cost of computing infrastructure, lowering the cost of cloud virtual machines you deliver because memory has been a significant portion of the I/O computing infrastructure and with the combination of Intel Optane and our software those costs can be lowered,” he said.

Memory Machine’s second capability is ZeroIOTM memory snapshots, which relies on the persistent qualities of Optane Persistent Memory to capture application states in memory.

“This allows [customers] to recover from failure gracefully and allows them to replicate [applications] more easily through our memory snapshot,” Fan said.

Fan said one of the benefits of working with Intel is that the chipmaker has acted like a reseller and has introduced MemVerge to companies that can take advantage of its software.

“They really almost serve as a channel for us to go to market, even though they don‘t officially resell our software, but they sort of lead us into these accounts,” he said.

One Intel channel partner that Intel introduced to MemVerge is World Wide Technology, a St. Louis, Mo.-based solution provider that is No. 9 in CRN’s Solution Provider 500 list.

Zach Splaingard, a server infrastructure technical solutions architect at WWT, said Optane Persistent Memory can enable new capabilities in servers that weren’t thought possible before. That includes virtualized desktop infrastructure, which is heavily dependent on large amounts of memory — something that can get pricy with traditional DRAM.

But Optane Persistent Memory in memory mode can make it much more economical. When using it in combination with DRAM versus only using DRAM, organizations can save 25 percent in data center costs, according to Splaingard.

“When you start to look at cost effectiveness of Optane to get large pools of memory, you can really up the density and get the most out of your capex spend,” he said.

Splaingard said he also sees a lot of promise in Optane Persistent Memory’s App Direct Mode and ISVs like MemVerge that can help applications take advantage of it.

“I see them really pushing the envelope with App Direct Mode, having a giant memory pool that serves as memory and persistent storage,” Splaingard said. “We’re really looking to leverage them for MongoDB and Splunk and other things like that, looking at it from a database performance perspective.”

WWT is working closely with Intel to test and sell Optane-based solutions at the company’s Advanced Technology Center, Splaingard said, and the company is seeing a “large uptick in interest” between conversations, proofs of concept and commercial deployments for applications ranging from VDI and databases to high-performance computing and AI.

“One thing that we’ve found constantly mentioned by customers is ATC allows us to do testing and figure out new opportunities and workloads better than they can do themselves,” he said. “It’s a big difference maker, and that’s why Intel has partnered with us to showcase Optane.”