Nutanix Weekly: Cloud Ready, Azure Steady

The rapid increase in data generation is making cloud storage an essential part of modern IT infrastructure. Driven by factors such as digitalization, the Internet of Things (IoT), social media, AI, and remote work, the volume of data created daily is staggering. 

Blog post: https://www.nutanix.com/blog/the-next-release-of-nutanix-unified-storage-is-here from Ujwal Datta & Alex Almeida

Host: Phil Sellers, XenTegra
Co-Host: Jirah Cox, Nutanix
Co-Host: Andy Greene, Nutanix
Co-Host: Chris Calhoun, Nutanix

Transcript

WEBVTT

1
00:00:03.250 --> 00:00:07.050
Philip Sellers: Hello, and welcome to another episode of Nutanix Weekly.

2
00:00:07.110 --> 00:00:25.349
Philip Sellers: one of the many podcasts coming out of Zentegra, this one being Nutanix Focused, and I'm your host, Phil Sellers. I am the Practice Director for Modern Data Center here at Zentegra, and … really happy that you're spending a little bit of time listening along.

3
00:00:25.390 --> 00:00:32.320
Philip Sellers: Can't do this alone, so I've brought my Mary … group of…

4
00:00:32.850 --> 00:00:50.499
Philip Sellers: co-hosts? I don't even know where to go with that one. So, anyway, introducing, our co-hosts for today, first up, Chris Calhoun, who is a solutions architect here at Zentagra, former Nutanix sales engineer.

5
00:00:50.520 --> 00:00:56.440
Philip Sellers: former customer of Nutanix, and former teacher. …

6
00:00:56.740 --> 00:01:01.570
Philip Sellers: Chris… Chris has had a few different lives, it seems like. Chris, how you doing?

7
00:01:01.880 --> 00:01:04.650
Chris Calhoun | XenTegra: Fantastic, as always. Thanks for asking.

8
00:01:04.790 --> 00:01:21.429
Philip Sellers: I mentioned that teacher thing because, honestly, I think it still colors how you interact with our customers today. You love teaching, you love sharing information, which is fantastic to have you here on the podcast sharing information every week.

9
00:01:21.670 --> 00:01:24.190
Chris Calhoun | XenTegra: Yep, definitely exciting. Appreciate it, buddy.

10
00:01:24.440 --> 00:01:28.840
Philip Sellers: Coach, too? Did I get that right? You were a coach at one time? Yes, absolutely.

11
00:01:28.840 --> 00:01:30.180
Chris Calhoun | XenTegra: Yeah, so….

12
00:01:30.180 --> 00:01:38.899
Philip Sellers: Coach-teacher makes a really great solutions architect, so, like that. We're also joined with Andy Green.

13
00:01:39.300 --> 00:01:45.349
Philip Sellers: Andy, also former Nutanix, former Cohesity employee, …

14
00:01:45.560 --> 00:01:51.619
Philip Sellers: Andy, … you've been here, what, a little over 6 months now?

15
00:01:52.200 --> 00:01:55.450
Andy Greene: We are rapidly approaching 9 months.

16
00:01:55.450 --> 00:01:56.350
Philip Sellers: Wow.

17
00:01:56.590 --> 00:02:08.910
Philip Sellers: And honestly, don't take it as an offense, but it feels like you've been here a lot longer. You're a rock star, and I definitely enjoy having you on the team.

18
00:02:08.910 --> 00:02:16.759
Philip Sellers: We were on a call earlier today, and Andy just absolutely rocked it out with one of our customers, and …

19
00:02:16.860 --> 00:02:20.940
Philip Sellers: just… Can't believe it's only been 9 months, honestly, so….

20
00:02:21.230 --> 00:02:27.210
Andy Greene: It is my pleasure. Honestly, I'm loving the team dynamic here at Zentegra. It's been a great fit for me.

21
00:02:28.550 --> 00:02:46.799
Philip Sellers: And, wrapping up the co-host, quadrants, we've got Jairah Cox. Jairah, is probably someone no one needs an introduction to. If you listen to this, you know Jairah, but he is the man, one of Nutanix's

22
00:02:46.800 --> 00:02:53.389
Philip Sellers: Field CTO, architects, … And just all-around nice guy.

23
00:02:53.470 --> 00:02:56.290
Philip Sellers: Jairah, it's great to have you on the podcast, too.

24
00:02:56.600 --> 00:03:00.690
Jirah Cox: Thank you, Philip. Just once, when a podcast host says.

25
00:03:00.690 --> 00:03:11.199
Philip Sellers: here, meet my bunch of Mary co-hosts. I want them to be co-hosts all named Mary. Today's not that day. Well, ….

26
00:03:11.700 --> 00:03:17.090
Jirah Cox: That seems like a long-term sort of commitment. Gotta have… gotta have goals.

27
00:03:17.090 --> 00:03:24.870
Philip Sellers: It's either gonna be a really hard recruiting thing, or it's gonna take, you know, 30 years or so for us to farm that out.

28
00:03:24.870 --> 00:03:27.959
Jirah Cox: One day, though, one day the stars will all align for all of our coasts.

29
00:03:27.960 --> 00:03:33.560
Philip Sellers: Well, on that note, Mary, we will, …

30
00:03:33.750 --> 00:03:51.280
Philip Sellers: We'll get started with a blog post today. More news coming out of Nutanix, and nobody's slowing down over there, and it's pretty evident, because the number of releases and announcements posted on Nutanix.com slash blog just continue

31
00:03:51.280 --> 00:03:56.979
Philip Sellers: To, to happen. And so, we had a tough time kind of talking

32
00:03:57.210 --> 00:04:04.139
Philip Sellers: pre-recording, figuring out what we wanted to talk about today. So, we settled on this, …

33
00:04:04.330 --> 00:04:21.620
Philip Sellers: here at Zentegra, we love Nutanix Unified Storage. We think it's one of the greatest parts of the portfolio, and so we've got news on Nutanix Unified Storage. So, Jaira, do you want to unwrap the news real quick and tell us what you've announced?

34
00:04:22.580 --> 00:04:29.159
Jirah Cox: A bunch of stuff. This is actually a very exciting article, with, a whole bunch of… a whole slew of.

35
00:04:29.230 --> 00:04:46.590
Jirah Cox: improvements for Ness here. The first one, right, right at the top is the introduction of Nutanix files running in the Azure environment, right? So there was some, some preview going on here, now it's here for real.

36
00:04:46.680 --> 00:05:02.979
Jirah Cox: And so now there's an ability to run ACNX files, not just on your on-prem clusters, not just in clusters that are in a colo data center or at the edge, but now also running in cloud on or without ACNX cloud clusters, right, or NC2 substrate. So this could be

37
00:05:03.050 --> 00:05:11.560
Jirah Cox: This could be the very first thing you ever deploy, is Nucinix files running in Azure there, without any HCI footprint whatsoever.

38
00:05:11.930 --> 00:05:12.520
Philip Sellers: Yeah.

39
00:05:12.830 --> 00:05:18.690
Philip Sellers: Yeah, that's exciting news. So, the name of the blog post today that we're talking about is Cloud Ready.

40
00:05:19.150 --> 00:05:29.969
Philip Sellers: Azure Steady. The next release of Nutanix Unified Storage is here. It's version 5.2. Blog was written by,

41
00:05:30.860 --> 00:05:35.380
Philip Sellers: Yoal, Dada, and Alex Alameda.

42
00:05:35.510 --> 00:05:36.680
Philip Sellers: …

43
00:05:36.990 --> 00:05:46.749
Philip Sellers: thanks for posting this so that we have something interesting to talk about. But, yeah, I mean, this is another

44
00:05:46.830 --> 00:05:59.510
Philip Sellers: kick forward for Nutanix Unified Storage. You know, we… we've talked about in the past what this looks like inside of AWS, so now it's coming to Azure, and…

45
00:06:00.220 --> 00:06:07.600
Philip Sellers: Let's recap for anyone listening real quick, Jairah. What does this mean that you don't need that

46
00:06:07.830 --> 00:06:11.440
Philip Sellers: Nutanix footprint, how is this actually running?

47
00:06:12.040 --> 00:06:26.720
Jirah Cox: Yeah, so, actually, Philip, you and I have had some shared customers in the past where the customers have wanted to add cloud to their Nutanix-shaped hybrid cloud. They've wanted to maybe use public cloud as a DR target.

48
00:06:26.720 --> 00:06:34.469
Jirah Cox: And one of the biggest components of their environment, that was for EUC specifically, was their profile storage, right? And so…

49
00:06:34.940 --> 00:06:39.310
Jirah Cox: That requirement to say, sure, you can run files anywhere.

50
00:06:39.690 --> 00:06:59.650
Jirah Cox: asterisk, anywhere you can run Nutanix, became kind of a long pole in the tent of their design, of, like, most of their cluster in the cloud running Nutanix was there just to run files and receive profile replication, right? Since it was an EUC environment, they had a small gold image template and a linked pool of desktops, they didn't need a lot of storage for VMs themselves.

51
00:06:59.880 --> 00:07:03.050
Jirah Cox: So with this kind of an optionality now that customers have.

52
00:07:03.100 --> 00:07:13.599
Jirah Cox: files, which is our software-defined, high-performance, scale-out, unstructured data storage, right? So, think file shares, think application data shares, think…

53
00:07:13.600 --> 00:07:24.209
Jirah Cox: home directories, think team drives, right? Everything you would do through a standard file server, hosted by Nutanix, can now run on a Nutanix cluster, maybe a great way to run it on-prem.

54
00:07:24.330 --> 00:07:39.930
Jirah Cox: can run it on a cluster in the cloud, or can run in the cloud without a cluster as well. So for, like, this customer, where the vast majority of their replication target would have been just files, well, now I can run just files on the cloud itself directly.

55
00:07:39.930 --> 00:07:49.380
Jirah Cox: running our Nintendix File Server VMs as native cloud VMs, and then using native cloud storage, both block and object.

56
00:07:49.620 --> 00:08:06.610
Jirah Cox: to be the actual capacity tier of that file server. And then, since I'm running mechanics files on both ends, I get that native replication simplified management outcome so that my data, right, my user directories are there when I want to actually need them, right? Like, for, like, a DR failover.

57
00:08:07.340 --> 00:08:25.160
Philip Sellers: Yeah, 100%. You know, we've talked to a number of customers about Nutanix Cloud clusters, or NC2 running inside of Azure, and it's not always the best fit. You know, Chris, you know, what kind of happens to customers when they start looking at NC2?

58
00:08:25.160 --> 00:08:31.379
Philip Sellers: And why does this NUS running directly on cloud give them a benefit?

59
00:08:32.419 --> 00:08:44.469
Chris Calhoun | XenTegra: I'll tell you this, from my experience in the past, I've had customers and, peers of mine from Nutanix trying to work on a solution for customers that wanted that

60
00:08:47.869 --> 00:09:05.399
Chris Calhoun | XenTegra: agility and flexibility to run ads in the cloud, but to Jairus' point, if the large majority of the data from on-prem that gets replicated to the cloud is file service storage, then that footprint is your

61
00:09:05.829 --> 00:09:07.239
Chris Calhoun | XenTegra: Longest pole.

62
00:09:07.589 --> 00:09:09.789
Chris Calhoun | XenTegra: Biggest hurdle, whatever you want to call it.

63
00:09:09.809 --> 00:09:23.509
Chris Calhoun | XenTegra: And that's something that's sometimes hard to overcome from just hardware. With some of the specific requirements around NC2 in Azure, you don't necessarily have

64
00:09:23.519 --> 00:09:30.419
Chris Calhoun | XenTegra: the flexibility to have 15, 20 different hardware models, all with expansive storage, so…

65
00:09:30.599 --> 00:09:37.459
Chris Calhoun | XenTegra: It became pretty expensive whenever, typically, a 4-6 node cluster on-prem would be…

66
00:09:37.579 --> 00:09:50.149
Chris Calhoun | XenTegra: 8, 10, 12, 14, to get the amount of storage that you need. So, that was a challenge from the past, and I think that, again, this goes to new 10X's willingness to listen

67
00:09:50.289 --> 00:10:04.859
Chris Calhoun | XenTegra: And even try to make the difficult technical things simpler. And that's where I think that we are now, with the flexibility of a native offering for the underpinning

68
00:10:04.949 --> 00:10:12.649
Chris Calhoun | XenTegra: That really helps and eliminates that hardware requirement and, avoids that compromise of having to

69
00:10:12.749 --> 00:10:19.219
Chris Calhoun | XenTegra: Overdue or create a larger footprint than is really, truly, truly needed.

70
00:10:20.120 --> 00:10:23.930
Philip Sellers: Yeah, you know, one of the things that strikes me, too, is, you know.

71
00:10:24.240 --> 00:10:28.280
Philip Sellers: for the use case Jair was talking about, with user data, things like that.

72
00:10:28.580 --> 00:10:39.840
Philip Sellers: We also need to protect that. We need DR. We need some way of making sure that that data is in a good spot. And so, you know.

73
00:10:40.310 --> 00:10:57.749
Philip Sellers: when you invest in a files platform, whether that's Nutanix files, or whether that's Azure Files, or, you know, even Azure NetApp files, you… you need the data portability tools, you need those features, and so Andy, …

74
00:10:57.850 --> 00:11:06.429
Philip Sellers: you know, what about SmartDR here? With… with this implementation, are you going to be able to do, …

75
00:11:06.700 --> 00:11:08.059
Philip Sellers: You know, the…

76
00:11:08.710 --> 00:11:16.560
Philip Sellers: The unified storage, sort of, move around data, that kind of thing within the cluster, or within files implementation?

77
00:11:17.240 --> 00:11:41.999
Andy Greene: Yeah, I think, you know, this is one of the areas where we're just gaining additional flexibility in how and where we run those shares. So Nutanix has that SmartDR feature. It's basically a disaster recovery feature that's built into Nutanix files, and it gives us that ability to replicate that share-level data between the different file server instances that are running on the various Nutanix clusters that we might be

78
00:11:42.000 --> 00:11:42.890
Andy Greene: managing.

79
00:11:42.890 --> 00:11:48.680
Andy Greene: Typically, you know, for DR purposes, we're doing this across, geographically, distinct sites.

80
00:11:48.710 --> 00:12:02.339
Andy Greene: And, you know, I think the key benefit here is it just gives us that ability to have a read-only copy out at those remote sites, but also very fast recovery times. We can quickly failover from one site to another in the event of an emergency.

81
00:12:03.340 --> 00:12:15.989
Philip Sellers: Yeah, and increasingly, our customers are choosing, you know, all sorts of different platforms. You know, we have customers that are still Citrix customers, but maybe they're dabbling in ABD,

82
00:12:16.070 --> 00:12:25.029
Philip Sellers: Maybe they're… they're looking at Omnissa, you know, running with ABD, and then they've got some… some things on-prem.

83
00:12:25.180 --> 00:12:37.860
Philip Sellers: But at the end of the day, they all have user profile data, and so, you know, with that, Chris, I'm kind of curious, I mean, is this a solution that we can pair with those

84
00:12:37.860 --> 00:12:46.390
Philip Sellers: orchestrated AVDs, whether it's Nerdio, or Citrix, or Omnissa, does this pair well with that?

85
00:12:46.610 --> 00:12:57.010
Chris Calhoun | XenTegra: Absolutely, and I really think that this gives customers the chance and the opportunity to test out a proven solution that's resilient.

86
00:12:57.010 --> 00:13:10.179
Chris Calhoun | XenTegra: One of the biggest things about files is that single namespace and the flexibility of running multiple file server VMs, that underlying foundation of Nutanix files. You know, in the past,

87
00:13:10.180 --> 00:13:19.689
Chris Calhoun | XenTegra: customers have had that Windows file share server and single instance, and that would be the gem of the most critical of

88
00:13:20.160 --> 00:13:32.089
Chris Calhoun | XenTegra: surprisingly, a single back-ended system. And that's one of the things that, with the flexibility of a hybrid model, yes, great, we'd love for you to,

89
00:13:32.460 --> 00:13:51.089
Chris Calhoun | XenTegra: continue to work with Citrix and AVD for those EUC workloads, but let's back-end it with a portable, flexible Nutanix file solution that provides a fantastic, resilient target for those profiles. I think that's going to be something that really lends itself to

90
00:13:51.090 --> 00:13:55.169
Chris Calhoun | XenTegra: Tools, capabilities, and yet portable flexibility.

91
00:13:55.630 --> 00:13:56.190
Philip Sellers: Yeah.

92
00:13:56.910 --> 00:14:04.340
Philip Sellers: So, Jairah, as Scott's questions, how does this work? Like, what's going on under the covers?

93
00:14:04.980 --> 00:14:18.659
Jirah Cox: So, under the covers, right, we actually are building, virtual machines on your behalf that are running the same OS, same application stack, right, that our FSVMs, file server VMs, run on-prem.

94
00:14:18.870 --> 00:14:30.320
Jirah Cox: It's still a scalable cluster, right? So, start with a minimum of usually 3 of those VMs, but then I can either attach more storage to those VMs to have them

95
00:14:30.320 --> 00:14:43.210
Jirah Cox: run as a logically larger NAS, or I can also scale out those VMs as well to then get even more throughput, drive more I.O. on the same dataset, or I can grow in both directions, right, if I choose to.

96
00:14:43.350 --> 00:14:49.640
Jirah Cox: whether I choose to, to scale out or scale up, what's neat here is that

97
00:14:49.650 --> 00:15:04.489
Jirah Cox: the system's going to automatically do intelligent data placement, and know, like… like we learned first two, what, 16 years ago, what's hot data, what's cold data, right? What should I do about it? And so that ability to blend data across both Azure Block Storage.

98
00:15:04.490 --> 00:15:17.549
Jirah Cox: read more expensive, or across Azure Blob Storage, more inexpensive. We can put the data where it can be the most effective to you as the user, but be the most economical on the monthly bill.

99
00:15:18.030 --> 00:15:38.020
Philip Sellers: I like the way you put that, you know, you blended two things that I wanted to talk about, because as soon as I saw, does it scale? my mind immediately went to the Does It Blend videos on YouTube, right? And so then you start talking about this blending it all together. So, it does both, right? You get the blend, and you get the scale.

100
00:15:38.020 --> 00:15:50.050
Jirah Cox: Yeah. And when most people hear about scaling and cloud, they're like, well, sure, of course you can scale. What does that do for my monthly bill, right? And so, scaling, but also being a good custodian of both your data and your dollars, whoa.

101
00:15:50.260 --> 00:15:54.120
Jirah Cox: That was awesome. Gonna write that down and reuse that later.

102
00:15:54.620 --> 00:15:57.060
Jirah Cox: It's just, you know, that's just part of the job.

103
00:15:57.760 --> 00:16:08.899
Philip Sellers: Yeah, I mean, that's the thing, right? We're really looking at this in terms of cost control, as much as we're looking at it from a performance standpoint. You know, we don't want it to tip over.

104
00:16:08.940 --> 00:16:19.590
Philip Sellers: Obviously, we want it to be performant if we're using it next to a highly intensive workload, like FSLogix or, you know, profile containerization.

105
00:16:19.680 --> 00:16:28.540
Philip Sellers: So, you know, there's all sorts of other use cases, too. So we're talking specifically around VDI, but…

106
00:16:28.680 --> 00:16:43.169
Philip Sellers: We, like, work with a ton of healthcare customers, and so, you know, PAC systems have a lot of data, and so scaling those out, moving those and refactoring those, you really need a, …

107
00:16:43.440 --> 00:16:51.429
Philip Sellers: powerful, scale-out architecture that you can count on. And so, … You know, it's…

108
00:16:51.710 --> 00:16:59.059
Philip Sellers: it's great that you have both options, right? You can go to the fastest tier, but you can also

109
00:16:59.200 --> 00:17:05.839
Philip Sellers: Make it a little more cost-conservative by… by tiering things off to blob storage.

110
00:17:07.960 --> 00:17:09.279
Philip Sellers: So, …

111
00:17:10.160 --> 00:17:21.889
Philip Sellers: you know, this… this is another drive towards something we've talked about in the past. I think it's called Project Beacon, right? That's the refactoring of your

112
00:17:21.940 --> 00:17:32.379
Philip Sellers: services on the Nutanix platform today to run independent of the Nutanix platform. So this is another example of that coming to fruition, right?

113
00:17:33.070 --> 00:17:46.019
Jirah Cox: I think it's that, as well as combined with sort of a… you know, as we help customers build… design and build their own cyber multicloud, it's understanding that change and migration

114
00:17:46.020 --> 00:18:01.379
Jirah Cox: is going to be a persistent element of that, right? Like, we can get it as most right as possible as we can today, but tomorrow, business is going to need different things out of it. So how do we help that stay flexible, stay agile, right, and skate to where the puck is going, kind of thing?

115
00:18:01.380 --> 00:18:02.000
Philip Sellers: Yeah.

116
00:18:02.390 --> 00:18:12.489
Philip Sellers: Andy, I remember we got a email, and I think we were both on it for one of your customers, and they were asking us a little bit about

117
00:18:12.540 --> 00:18:25.549
Philip Sellers: you know, how do we tell when this Windows file server's at capacity? And, you know, what tools do we have, and how can we, …

118
00:18:25.720 --> 00:18:26.850
Philip Sellers: do…

119
00:18:27.100 --> 00:18:38.709
Philip Sellers: any sort of performance testing and stuff like that with it. It's not easy to do on Windows, is it? And what's the benefit here, I guess, for customers when they're looking at Nutanix files?

120
00:18:39.630 --> 00:18:54.169
Andy Greene: Yeah, well, you know, Nutanix Files has always been just incredibly scalable from a performance perspective. Jairra touched on that earlier. And we're also extending that unified namespace across, you know, many different,

121
00:18:54.170 --> 00:19:07.970
Andy Greene: Nutanix platforms. One of the updates that I've seen as part of this project is that now we'll also have the ability to extend that unified namespace across some non-Nutanix systems as well.

122
00:19:09.010 --> 00:19:21.070
Andy Greene: So, it gives us that seamless way to bring external file shares into that single namespace, and it also gives us some flexibility around phased migration approaches.

123
00:19:21.580 --> 00:19:25.419
Philip Sellers: Yeah, and see, that's all a benefit to the customer. Reducing risk.

124
00:19:25.660 --> 00:19:28.939
Philip Sellers: That risk of, you know, something hitting a bottleneck.

125
00:19:29.270 --> 00:19:32.150
Philip Sellers: And giving you an alternate path to move.

126
00:19:32.920 --> 00:19:33.919
Philip Sellers: I love that.

127
00:19:34.690 --> 00:19:41.369
Philip Sellers: … You know, when we talk to customers, a lot of this…

128
00:19:42.580 --> 00:19:58.920
Philip Sellers: concern about risk, migration, those kinds of things come up, Chris. You know, with something like Nutanix files, I've got to think this may be the last time you have to do a migration?

129
00:19:59.870 --> 00:20:13.410
Chris Calhoun | XenTegra: Yeah, definitely. That's one of the biggest things I remember from my days, of course, as a customer is, those EOL periods where you had to swap out hardware, repave over everything, migrate data.

130
00:20:13.500 --> 00:20:27.129
Chris Calhoun | XenTegra: all the like. Well, with Nutanix, obviously, as you bring in both either on-prem infrastructure, and then even age that out, it's a matter of just, of course, adding an additional node. And in this case.

131
00:20:27.130 --> 00:20:43.460
Chris Calhoun | XenTegra: If you can flex to the cloud for portability and have that capability, your data is pretty much not localized to one data center, but flexible enough for you to access it from anywhere. And I think that that's really streamlining the business of the future as far as

132
00:20:43.460 --> 00:20:56.640
Chris Calhoun | XenTegra: making your data portable, and not locking it into that one server under somebody's desk, or that one piece of shiny hardware in the data center that runs absolutely everything. It's being

133
00:20:56.970 --> 00:21:04.919
Chris Calhoun | XenTegra: Agile enough to have resilient systems that are flexible and that will allow your data to flow from

134
00:21:04.960 --> 00:21:19.829
Chris Calhoun | XenTegra: both file shares on-prem to the cloud, and vice versa. So I think that that's really where Nutenix flexibility lends itself, and is the glue for both, hybrid on-prem and, cloud-based environments.

135
00:21:20.750 --> 00:21:33.440
Philip Sellers: Yeah, I know the last time I did a large-scale sort of files, multi-terabyte, I want to say nearly 100 terabyte files migration, it took several weeks, and that was doing…

136
00:21:33.440 --> 00:21:42.030
Philip Sellers: you know, I don't know, 10 lanes of robocopies on different directories in tandem, and, you know, we had all these sort of…

137
00:21:42.880 --> 00:21:53.830
Philip Sellers: Scripted ways to get around, you know, throughput or bottlenecks to try and make it happen, and it still took weeks or months to make those transitions happen.

138
00:21:54.080 --> 00:22:01.050
Philip Sellers: Being able to do it in a way that preserves user access throughout that transition, you know, the external

139
00:22:01.360 --> 00:22:07.589
Philip Sellers: or the logical namespace, I think those are huge, huge benefits

140
00:22:07.720 --> 00:22:21.939
Philip Sellers: that are now delivered. I know I talked to a customer recently where external namespaces came up as a way of unifying pools of storage within their Nutanix clusters

141
00:22:21.940 --> 00:22:36.780
Philip Sellers: where they didn't have enough total storage, but by tapping into A, B, and C, now they have enough total storage, and so there's some huge benefits there, I think, for, you know, captive storage capacity.

142
00:22:39.550 --> 00:22:43.400
Philip Sellers: The, … Last thing I would say…

143
00:22:43.670 --> 00:22:53.700
Philip Sellers: Jairret, you and I did a podcast talking about AI and ML workloads, so we've talked about user profiles, we've talked about PAX data, imaging data.

144
00:22:53.820 --> 00:23:01.850
Philip Sellers: There's a whole nother workload that we haven't talked about, AI and ML, that uses unstructured data

145
00:23:02.030 --> 00:23:09.009
Philip Sellers: And we had a fairly big announcement. Nutanix Unified Storage performs like a rock star.

146
00:23:09.240 --> 00:23:14.420
Philip Sellers: when you're running underneath those kinds of workloads. So, do we get that here, too?

147
00:23:14.950 --> 00:23:28.690
Jirah Cox: Yeah, believe it or not, it almost is a… it's a funny way to think about it, right? We were, top of the heap, very proud of… of our results, right, for the ML Perf benchmark, several blog posts and several episodes of the podcast ago.

148
00:23:28.750 --> 00:23:45.939
Jirah Cox: But I guess that wasn't good enough for us, right? So, not wanting to… not wanting to sit still on that. With the enhancements now that come with Files 5.2, powered by SRIOV of the Nutanix platform, we can now offer storage also over NFS over RDMA.

149
00:23:45.940 --> 00:23:57.839
Jirah Cox: which has dramatic reductions, in, latency access, allows us to do memory-to-memory data transfers between storage and GPU clients for GPU direct workloads.

150
00:23:57.840 --> 00:24:18.519
Jirah Cox: Right? We can now feed those… I saw them as referred to as, AI accelerators. They're really… they're GPUs. That's what they are, they're GPUs. We can now feed those GPUs, just vast amounts of data even more quickly, right? So, like, streamlining, that high-speed data path between the GPUs that want to do the inference and do the analytics.

151
00:24:18.520 --> 00:24:25.199
Jirah Cox: and the platform that hosts the data, us, hello, that lets you do all of that AI work that you want to do.

152
00:24:25.200 --> 00:24:27.270
Jirah Cox: even much more rapidly, right? Which…

153
00:24:27.270 --> 00:24:41.410
Jirah Cox: you know, when you're buying these assets, the faster you can do it, A, better for the business, but B, I can get onto my next job task, right? So then they can stay… they can do even more work for me and be… and lower that effective cost for, each AI, you know, token, if you will.

154
00:24:41.700 --> 00:24:51.970
Philip Sellers: Yeah. Well, and so what I'm hearing you say is basically it's like Pac-Man for GPUs, you're just basically letting the GPU eat faster, so….

155
00:24:51.970 --> 00:25:03.530
Jirah Cox: Right. So yeah, that is a fair way to put it. Files 5.2, it is that big dot that makes all the ghosts blink, and then you can go even more. That's a very apt analogy.

156
00:25:04.570 --> 00:25:18.049
Philip Sellers: Well, you know, this has been, a long time coming. You know, Nutanix Files is a mature platform. The version number 5.2 lets you represent that it is absolutely

157
00:25:18.190 --> 00:25:24.249
Philip Sellers: Growing up, you know, one of the new things also in this release

158
00:25:24.280 --> 00:25:44.110
Philip Sellers: metro cluster, so the ability to do synchronous, and so this was something we weren't able to do. It comes into place a lot of times when we're doing, you know, sort of split workloads, VDI on both sides, active-active. So this is a really great additional feature, too.

159
00:25:44.110 --> 00:25:44.760
Jirah Cox: No.

160
00:25:44.760 --> 00:25:51.079
Philip Sellers: You know, that real-time mirroring across geographic boundaries is also huge.

161
00:25:51.150 --> 00:26:07.559
Philip Sellers: Typically we've had to include a third-party software to be able to achieve this. That's great that this is now coming. What… do you know if there are any, restrictions, requirements around being able to do Metro with files?

162
00:26:08.150 --> 00:26:24.660
Jirah Cox: The biggest one is laws of physics, still apply. So, you know, must be… must have two data centers that are close enough to each other to do it, right? That's usually the biggest one I see among, my customers that I consult with. But, it's cool, right? It's IT. Traditionally, we've just simply bought two of everything.

163
00:26:24.660 --> 00:26:34.170
Jirah Cox: Right? And put them some distance apart, and then done some amount of work, maybe zero, maybe more work, to be able to do failovers across those two environments, right?

164
00:26:34.330 --> 00:26:41.629
Jirah Cox: given what we can do with modern software, modern hardware, right, this metro deployment of, like, look, I'm still gonna buy two of these things.

165
00:26:41.820 --> 00:26:50.470
Jirah Cox: But I effectively want to do as little work as possible to do a failover, right? If something goes bumped, if a truck hits the wrong pole at 3 in the morning.

166
00:26:50.790 --> 00:27:01.220
Jirah Cox: can everything just fail over? Yeah, you know, a couple minutes while VMs boot back up, but can everything just fail over, and no humans get out of bed, nobody gets on a VPN on their laptop.

167
00:27:01.220 --> 00:27:13.229
Jirah Cox: just make it all work. And we've had that now for VMs for a while now on Metro AHV, but it's so great to see files come to that party now, too, right? Because now my systems that rely on those file shares

168
00:27:13.230 --> 00:27:21.700
Jirah Cox: get that auto-recovery outcome, right? My desktops get that auto-recovery outcome that they were using it for in the first place. So, wildly exciting, glad to see it.

169
00:27:22.030 --> 00:27:23.340
Philip Sellers: Yeah, absolutely.

170
00:27:23.570 --> 00:27:24.530
Philip Sellers: So…

171
00:27:24.750 --> 00:27:40.860
Philip Sellers: We're gonna wrap things up there. I think, 5.2 is an exciting release. It's something we rely on with our customers here at Zentegra. If you're not using files, did you know that you've got actual entitlement to it? Every single cluster

172
00:27:40.860 --> 00:27:46.350
Philip Sellers: of Nutanix, comes with a terabyte of files entirely, maybe two.

173
00:27:46.410 --> 00:28:02.009
Philip Sellers: I get it confused. You get entitlement for both files and objects. If you've never tested it out, it's a really compelling thing you can add to your cluster. For all our DRAS customers, we recommend that they adopt this, because it gives superior

174
00:28:02.010 --> 00:28:12.019
Philip Sellers: DR capabilities when it comes to unstructured data. And so, with that, I'm gonna say thanks to Chris, Andy, and Jairah. Appreciate you guys.

175
00:28:12.040 --> 00:28:21.559
Philip Sellers: meeting with me, talking about it, and thanks everybody who's listening. We will see you on the next podcast. Have a great day.