00:09
Brady Kachapuram: Hey Hello everyone, welcome to another edition of Solutions In 6 series. Super excited to bring our special guest today introduce our special guest today, our virtualization guru ransomware guru is surviving who ECP Eric 10x vExpert. Let me get it I get 10x vExpert, Andrew
00:33
Andrew like I remember you doing a lot of storage architecture back in the day, right? Do you want to go like walk us to do that and some pointers to friends here. Andrew Miller: Absolutely happy to I, I definitely appreciate it. I put it in there the vExpert community over the last
00:51
10 years, it's been through a lot of changes in my career. And it's so many good folks there. But to your point, a lot of work over the years around VMware storage and VMware architecture storage across a couple different platforms, it's been more than fair, actually. So when I think about that, there
01:05
is a bunch of stuff that comes to mind that I used to plan around, there's some there's some big high level categories to be real. Sometimes we think about optimizing for costs, we think about optimizing for availability, we think about optimizing for performance, those are all legitimate things.
01:19
And I'll I'll get out of the way there for just a second. So you can see those cost ability and performance. But then there's all these other things that we dive into a layer down to we're thinking about maybe data reduction, deduplication and compression on or off. What's the performance impact of that
01:33
stuff? If I'm thinking about, you know, what level of protection do I need from a RAID level perspective? Or even overall resiliency, 5, 6, 7, nines based on the application? Do I need encryption? Or not? Or can I change my mind on that down the road? Is there a performance impact when I do it?
01:49
What's even the latency profile, you know, sub millisecond, 510 milliseconds? And even not just what's the latency profile? But what's the predictability of that latency isn't going to be consistent? Sometimes I look at you know, I've got different tiers of storage, and will things move around and maybe
02:04
unpredictable ways? And then of course, last but not least, noisy neighbor? I know, do I have no colocated? VMs. They're gonna step on each other. And if I'm doing upgrades, what is performance look like during that? How do I have to plan for unscheduled downtime? And so that's, it's almost like this
02:20
laundry list of stuff. There's a lot of stuff to think for, think about and plan for when it comes into VMware storage design. Brady Kachapuram: That's great. I know, you've talked a lot about the challenges, right? Can you take us to that orange? What does this look like with flash arrays?
02:40
Andrew Miller: Yep. I think about that with Flash, right? And we're making a bold claim here, you know, or, but putting it out there kind of, you know, slasher, the best VMware stores, the reason, and so much of what I've done in the past is what goes in here, think about nothing to tune in not having
02:54
trade offs being ideal for all workloads, it's like it says on the slide, or if we're gonna have fun with memes, current memes, you know, so, you know, ancient Star Wars movies, you know, if you're architecting storage by VMware environment. And, you know, maybe if you're not using FlashArray, someone
03:08
might look at you a little bit funny, you know, kind of thing. But why is that though, let's take some of those same categories and kind of walk a step down. So think about cost optimized, that's always a big category. So from a Pure standpoint, there's automatic data reduction. This is both
03:23
inline and post process deduplication. And compression, we don't have to think about configuring RAID levels, it's all it's n plus two, we don't think about aggregates and storage pools and all this kind of stuff. updates are delivered via the cloud. And even as simplicity of upgrades
03:38
standpoint, that doesn't have performance impact. And then we think about, you know, what's the same multipaths, that's gets down in the weeds, but actually optimized around the native MPI o approach, round robin equals one for ESX, multi pathing. Continuing on, so we think about optimizing for availability. So
03:54
Pure, not only has a very highly available architecture, to even do upgrades, there's no downtime. But we're now at the point where we can talk about having seven nines of uptime. From a fleet wide perspective, I want to have a little bit of fun with math, seven nines is actually three and a half
04:08
seconds. What that actually means is that most of our customers have no downtime, there's something to do, but it only pulls the average down enough that seven nines and we're not taking the cheat card there of taking out plan down to just for Plan time for either planned upgrades or damages or
04:23
datacenter moves. Always n plus two data protection, encryption is always on period, you can't turn it off, we actually use it in some ways for Wear leveling, that actually helps out in so many ways. And then of course, we think about performance optimized so you know, I'll get out of the way here a little bit
04:38
for a second, consistently sub millisecond latency on FlashArray x. If we want to have a different level, we can talk about FlashArray see the different price point that has great latency as well. QoS is always on to help with noisy neighbors and know calling it out again, because it matters
04:52
100% performance through failures and maintenance. Just very cool. There's a lot of hard architecture there. And it's neat to sometimes go into how MPI V works, the active standby controller architecture even go some back to the core architectural decisions that we made with Pure going back to the
05:07
very founding of the company. Brady Kachapuram: These are all cool and relic. There's some customer related points here as well on the availability and uptime. But I mean, these are from Pure vantage point, since we are talking about VMware here, right? What does VMware
05:23
have to say about this, Andrew Miller: So lets kind of f ip it around. So I can say c ol things about Pure hopefully, I sound kind of credible, as I s id, but what's actually neat here is that, you know, if we look at our relationship w th with VMware as Pure, we have
05:37
esigned partnerships in multipl areas. You know, we're design partner on vVols, with VMwar Cloud Foundation, NVMe over abrics pushing that forwa d as a transport layer of VMware cloud-native solutions, great w rk we do together and eve Site Recovery Manager, you know, actually developing that o
05:53
t and doing work with VMware w ere a lot of the development i done for Site Recovery Manager n top of Pure products and makin sure compatibility. Then go on step further. You know, we think about vVols, I mentioned that they're, you know, there's this gentleman named Lee Cas ell. He's a VP of Marketing for
06:07
loud Platform. Bu so he kn ws a little bit about vVols. nd you know, he's on record nd saying, literally a quote h re that pure is the number ne platform for vVols deploye . It's really cool. Of course, f you want to try this stuff ou , I think you've got a coup e thoughts ther
06:23
Brady Kachapuram: Oh, yeah. So I mean, if you want to test all the things that Andrew just said and want to decide if Flash rray is the best storage for VMware, just pop onto our website, Pure storage.com and click on that big train of logo on the website. And thank you. Thank you so much, Andrew, for
06:44
talking about VMware and pulling up your teknicks V expert. expertise are appreciated. Andrew Miller: Thanks for it. Thanks for listening.
