Dell EMC VxRail and Intel Optane - Is Pmem HCI Ready?
Transcript
(smooth jazz music) >> Hey is Keith Townsend principle of The CTO Advisor. Thank you again, Intel for sponsoring this series of videos with Dell VxRail team. Today we have on the other side of the camera, virtually, is Bill Leslie senior manager Dell technical marketing for VxRail. Bill, welcome back to the program. >> Thanks Steve, appreciate it. >> So in this conversation, we're going to dive into Intel Optane. We've heard a lot about Intel Optane at the recent Best of VMworld.
They won for one of the most innovative technologies. But, the problem with innovative technology is that adoption can be relatively slow. So Bill, VxRail, talk to us high level the relationship between VxRail and Intel Optane. >> Yeah so, you know, we've got a great relationship with Intel. We work with Intel constantly on bringing out new technologies. Harnessing the power of those technologies and bringing them out to the places where customers are going to get the benefit from it. And I think you're right.
There's this early side of the adoption curve right now where there's some hype around it. But we've got some great evidence in terms of the types of performance and actual real world tangible benefits that customers can take advantage of today with VxRail and the Optane technologies. >> So my audience already knows I'm heavy into SAP HANA and I'm familiar with Intel Optane from that world. But when I think about the VxRail platform, I think about replatforming my entire data center around VMware Cloud Foundation, all the management capabilities of VxRail.
I don't necessarily think about these high-performing applications. I think about, you know, what this just... The general purpose pod, help me understand where you guys are seeing benefit for putting, you know, these race car engines of memory solutions inside of VxRail. >> Well, as soon as you hit it with SAP HANA, this is a use case that we've been focusing on over the last few quarters. Now we've got a four-socket VxRail platform or P580N and it's an all PMM platform that is certified with SAP HANA.
And for exactly the cases where you're talking about where SAP has gone through the retooling, the refractory to take advantage of these types of devices in those architectures. So a single Optane PMM in app direct mode where you're letting SAP take care of all of the replication, all of the protection levels, that single device can deliver a million AIOps. That's transformational types of gains and performance that can be taken advantage of right now with VxRail. >> So what's the difference between me doing that and just simply loading the VxRail system with RAM?
I can put a few terabytes of RAM into my VxRail. Is a standard second generation Intel Scalable Platform. 5 terabytes of RAM per socket so? >> Yeah, great question. Because really what you're changing is how that persistent memory is used. You're actually, in the app direct mode, you're using it as a persistent tier of storage rather than as memory. So you're going directly from the VM to the PMM card. You're bypassing a whole lot of IO and different operations that traditionally need to take place.
I know we don't need to go into all the details on those. But by actually going direct to the device, you're able to cut out all kinds of networking challenges as well as all of the additional operations there and getting a million AIOps per device. A little bit, this is one of those cases where a little bit it doesn't just go a long way, it goes far, far away in terms of the performance spectrum and really changes the game. You can actually load up to 12 of these in a single node, 12 million plus AIOps in a single node of VxRail.
It's rather powerful in terms of performance for things like in memory databases. >> Well, so as I'm thinking about 12 million AIOps and a single two use server that's my... My mind is kind of catching up to this concept. The advantages can't only be in memory servers. Like that's a lot of AIOps and for most of my database needs that's going to be, storage is no longer the bottleneck. >> That's right. >> Are there any other examples like sequel or where people are using this as just regular storage level memory?
>> So the other use case with Optane is the memory mode. And this was something that we just recently are bringing out with the VxRail. And think of memory mode as a memory tiering type of use case. So you're still going to use traditional DRAM. And then within vSAN, you're going to use the Optane memory mode as this larger pool of memory at a lower cost point. So you can work on some workload density optimizations within that. And it still delivers near equal performance as you would with just traditional DRAM.
So it's a completely different use case than the app direct, but now it's moving, that is a memory class, true memory type of use case with the Optane persistent memory. >> So if I'm understanding this correctly there's two inverse examples. The first example is where I'm taking an in memory database and using this storage feature of Intel Optane PMM to basically cache... My inner memory rolls into this permanent storage that's super fast, bypasses all of the fluff that we built over the years for storage.
So that I'm one, spending less money on my in memory database size 'cause I have less physical storage but I'm not sacrificing much in performance. And then on the inverse, if I have a traditional VM environment, and we all know that memory is almost always the bottleneck for most enterprise VMware deployments, I can load up systems with more memory then leverage vSAN to... Basically provide this cheaper caching layer? >> That's right. The memory mode does provide that additional memory caching layer of memory, which is really, really beneficial not only to traditional databases but almost any workload for that matter.
VDI environments tend to be very memory hungry as well. Great way to pack even more lower price point memory tier storage into VxRail for those VDI environments. >> So we've talked a lot about really big workloads, really big systems four-socket systems. Most of us don't have four sockets. >> No. >> Does this work on a traditional two socket system where most of my customers kind of play around? >> It does, Keith. And it even works on some of our one use systems, right?
So depending on whatever your footprint is, your workload requirements, we can satisfy those. 0 hours with our VxRail system. And that's not necessarily using the Intel Optane. That's just some performance that our customers can very easily upgrade to. There's 36 different versions of the VxRail software that they can directly upgrade from. For some reason, they haven't been keeping up to date with us, making it very simple for them to just click the one-click upgrade within VxRail and get the latest and greatest and untap that performance.
It's just sitting there ready for them to utilize. >> So Bill, last question, the combination of VxRail on DELL hardware. I understand the value, I get VMware Cloud Foundation, I get these ability to manage the devices off to the edge. We talked about that in different conversations. But tell me... Why Optane and VxRail, like what's the advantage of this management concept of VxRail and Optane. What am I getting by buying those two together bottom line? >> You're really getting with the VxRail the simplicity of management and operations.
All of our functionality is directly built into the vCenter plug-in for VxRail. There's not a separate tool that you've got to learn, a separate UI, different net, new workflows that are radically different or us trying to duplicate what VMware does best with their own software. You're using native VMware tool sets with that. What we're doing is we're taking that system level approach and tying everything from vSphere, vSAN, down through VxRail HCI system software into one integrated experience. So that if a customer wants to add Intel Optane as a cache drop, very easily done.
It can get much lower, more predictable performance. They want to utilize it to get the screaming-fast performance for one of those in-memory databases, you probably only need a relatively small number of persistent memory devices to meet the performance for your in-memory workloads. Or if you're trying to get that workload density, lower that price point of memory, you can use the Optane Intel technologies for persistent memory using memory mode. It's really a win across the board in terms of utilizing these different Intel technologies so that customers can get the advantage where and when they need it.
Not just us pushing it down and saying, okay, you've got to buy this future proof your data center. The time is now. These technologies are being taken advantage of in VxRail. And in our customers are definitely thanking us for it. >> So Bill Leslie, thank you again for sitting in the hot seat. I know I don't ask the most friendliest questions even though I'm doing it with a smile. That's it for this episode of The CTO Advisor. com, me on Twitter at CTOAdvisor.
Talk to you next CTO dose of Keith. (smooth jazz music)
