49: Nutanix Weekly: Nutanix Files 4.0 New Features: Performance Improvements

Apr 12, 2022

The Nutanix Files™storage solution introduces new enhancements in its newest 4.0 release that improve client performance, enhance file server compute efficiency, and deliver a consistent experience in mixed environments.  We’ve also done some scalability testing we’re excited to share.

Host: Andy Whiteside
Co-host: Havey Green
Co-host: Jirah Cox


00:00:02.760 –> 00:00:18.210
Andy Whiteside: Everyone and welcome to episode 49 of new techniques weekly i’m your host Andy whiteside i’m coming to you on location from Orlando Florida where my family is it does the animal kingdom, and I am in the hotel room talking to these two guys hey guys how’s it going.

00:00:18.840 –> 00:00:19.710
Harvey Green: pretty good.

00:00:19.770 –> 00:00:21.960
Jirah Cox: pretty good Hello from not as the world.

00:00:22.500 –> 00:00:22.890

00:00:25.260 –> 00:00:25.500
Andy Whiteside: But.

00:00:25.560 –> 00:00:28.470
Harvey Green: I was gonna say i’ll be kind of living that life later this week.

00:00:29.730 –> 00:00:35.160
Harvey Green: Take the family down the floor, and so they can go how spring break so they can go have spring break.

00:00:36.330 –> 00:00:40.470
Andy Whiteside: we’ll have to have to mute out this part so when my wife listens to the podcast later.

00:00:43.890 –> 00:00:52.980
Andy Whiteside: You know couple days and apart couple days, working from the hotel room that’s probably a decent balance because well you know what happens when you’re out.

00:00:53.700 –> 00:01:08.430
Andy Whiteside: doing the family stuff all the time, yes, yes, no i’m just fortunate to have a healthy family where we can go enjoy stuff like this stuff and so we’re going to show must go on right we’ve been on i’ve been at the park for three days it’s time to time to work a little bit.

00:01:11.220 –> 00:01:11.820
Harvey Green: Nice.

00:01:12.450 –> 00:01:17.160
Andy Whiteside: So good Harvey green almost almost got you Harvey Cox and jive again.

00:01:20.010 –> 00:01:20.310
Andy Whiteside: yeah.

00:01:20.520 –> 00:01:21.720
Jirah Cox: Well, we can powers combined.

00:01:22.260 –> 00:01:26.490
Andy Whiteside: Right that’s one way to look at it wonder twin powers activate.

00:01:29.700 –> 00:01:38.250
Andy Whiteside: All right, so i’m rv and gyro V we actually discovered something a few minutes ago that I did not realize actually I probably realize it a couple days ago.

00:01:38.580 –> 00:01:44.010
Andy Whiteside: we’ve been covering the details next corporate blogs, some of them can be marketing and fluffy.

00:01:44.310 –> 00:01:53.550
Andy Whiteside: And we decided we needed some deeper content, so we went after the Community blogs and man, did we find content over here so we’re going to be mixing those in from time to time.

00:01:54.060 –> 00:02:00.900
Andy Whiteside: job or any any insight on the new tannic community and and how that has been beneficial to new tannic the company.

00:02:02.190 –> 00:02:05.280
Jirah Cox: here for sure it’s a it’s our host tech Community called next and.

00:02:06.540 –> 00:02:19.560
Jirah Cox: Where you’re new to a couple of zero right that our staff even blog here as well, right, so the Community can write their own stuff our tech folks can join them as well and that’s a fantastic place to get them gets more content.

00:02:21.450 –> 00:02:24.900
Andy Whiteside: And I guess the the marketing guys do it from afar.

00:02:27.270 –> 00:02:33.900
Jirah Cox: A lot of these people i’ve seen here actually are, in our tech marketing so they’re they’re really smart guys.

00:02:34.290 –> 00:02:34.890
Jirah Cox: Smart folks.

00:02:35.100 –> 00:02:43.800
Andy Whiteside: Well, what I meant by that was your traditional marketing folks but yeah tech marketing a lot of those are the cream of the crop people that rose up through the world and decided to take a different path.

00:02:44.400 –> 00:02:44.970
Jirah Cox: yeah absolutely.

00:02:46.080 –> 00:02:56.760
Andy Whiteside: You know rv we you know, several several several areas of the country, whereas integrity is physically located, we need to probably get involved more in the new tannic communities that something we should probably take a.

00:02:57.390 –> 00:03:03.300
Andy Whiteside: More active interest in I know it’s happening pandemic probably slowed a lot of it down, but hopefully it’s going to come back around.

00:03:03.900 –> 00:03:14.130
Harvey Green: yeah absolutely definitely agree with that, I mean it it only helps for for everybody in the Community, when when you actually you know put something in it so.

00:03:15.210 –> 00:03:20.490
Harvey Green: The more we can do there, the better it will be, and I mean the same for the Community to us to stop.

00:03:22.050 –> 00:03:31.440
Andy Whiteside: So the blog title is new tannic files for auto new features performance improvements it’s by Mr mcgee who, that is.

00:03:31.740 –> 00:03:35.040
Jirah Cox: yeah might be game he’s on our files to marketing team okay.

00:03:35.700 –> 00:03:50.100
Andy Whiteside: So, and I should apologize Now for my speak Maya microphone if it’s not great i’m improvising here on the road, and I did her that could not the pack anything more than the head, too, so not a single headset went with me, which in hindsight might have been a bad idea.

00:03:53.760 –> 00:03:54.120
Harvey Green: Okay.

00:03:55.020 –> 00:03:55.590
Harvey Green: you’re okay.

00:03:55.980 –> 00:04:05.220
Andy Whiteside: I got all my clothes in my technology to this one bag and then my wife walks out with two suitcases and the whole goal was not to check luggage, so I had to sneak onto the plane, with two bags.

00:04:07.860 –> 00:04:10.920
Andy Whiteside: got I got hyper converged only to get.

00:04:12.000 –> 00:04:15.000
Andy Whiteside: Three tier back when it was founded on the point that makes sense.

00:04:17.490 –> 00:04:18.450
Andy Whiteside: To to release.

00:04:19.020 –> 00:04:28.320
Andy Whiteside: Right alright, so this this blog Mike Magee there’s four topics that he simply brings up and I want to bring up something real quick, you know.

00:04:28.710 –> 00:04:32.490
Andy Whiteside: We do a lot with hyper converges integrity about all we do is hyper converged.

00:04:32.910 –> 00:04:41.430
Andy Whiteside: And there’s lots of things around the io of the data and the redundancy of it and the ability to make a redundant and multiple data centers.

00:04:41.760 –> 00:04:45.300
Andy Whiteside: There are multiple clusters even into the cloud that’s that’s all extremely important.

00:04:45.780 –> 00:04:53.370
Andy Whiteside: And you can do that with other technologies, but what you can’t do with other technologies, other than new tonics really is this files piece.

00:04:53.820 –> 00:05:02.310
Andy Whiteside: And that is super important for workloads, one of which we spent a lot of times with which is the end user computing vdi virtual desktop virtual APP space.

00:05:02.760 –> 00:05:12.300
Andy Whiteside: You know, probably number one complaint in our space for forever was slow login times and people really just got used to it, they really just got to a point where they just accepted it.

00:05:12.810 –> 00:05:20.430
Andy Whiteside: And didn’t have to be that way and files and having that data local and now with fs logics in the mounting of those containers real time.

00:05:20.940 –> 00:05:30.360
Andy Whiteside: A lot of it’s been improved by improving the technology, improving the architecture and mechanics is just absolutely perfect for for the files piece of a hyper converged solution.

00:05:32.790 –> 00:05:42.780
Andy Whiteside: So the the four bullets that are actually called out here is improved synchronous random right performance that was over my head that was going to get a gyro cpu scheduling with FFS.

00:05:43.530 –> 00:05:56.670
Andy Whiteside: flows flow control and new tonics volumes and pushing the performance envelope so we’ll start with the first one driver this improving synchronous random right performance, how is that elevating what files can do for a customer.

00:05:58.170 –> 00:06:06.060
Jirah Cox: So first on the on the back end of the file system as you’re writing user data to disk right within the files.

00:06:06.630 –> 00:06:11.580
Jirah Cox: file system because that’s its own file system runs back in the back of the behind the scenes there.

00:06:12.570 –> 00:06:23.100
Jirah Cox: The way that it can use the fastest tier of storage, like, for example, opt in to do the logging as to what rights occur that translates into more end user receivable.

00:06:23.910 –> 00:06:33.330
Jirah Cox: Right performance, so the way that we can intelligently use we’re just going to all benefit of files in general right is that ability to intelligently place the hottest data, where it needs to go for the fastest.

00:06:33.780 –> 00:06:39.930
Jirah Cox: and user experience so bringing that out to devise a logs of the file system itself on the back end.

00:06:40.440 –> 00:06:49.110
Andy Whiteside: You know I never really thought about this, but so if i’m working within a vm and i’m writing to the virtual disk i’m writing through the new tannic file system.

00:06:50.220 –> 00:07:00.450
Andy Whiteside: And that’s happening that’s that single layer right words it’s writing to disk and then this writing somewhere else to disk and at that point, it officially tells the system it’s been written right, if I have seen rf have one.

00:07:01.020 –> 00:07:03.330
Jirah Cox: So we do that, for our have to yeah.

00:07:03.360 –> 00:07:07.260
Jirah Cox: Right two copies but um but yeah so we do that for all data.

00:07:07.770 –> 00:07:17.760
Jirah Cox: Everywhere right, but in this case also it or first two files itself runs the virtual fs vm that are the front end and then he also uses block storage on the back end and it uses.

00:07:18.120 –> 00:07:25.620
Jirah Cox: A file system on top of those block devices to actually store the nfl SMB data so there’s a there’s a separate level of walking that goes on there as well.

00:07:25.890 –> 00:07:32.310
Andy Whiteside: yeah so there’s an extra system in place here, and this is really giving performance underneath that system so that.

00:07:32.730 –> 00:07:41.580
Andy Whiteside: It becomes less and less of a constraint, even though probably wasn’t much of a constraint before when you’re talking about heavy rights reads or writes in this case every little bit helps.

00:07:41.820 –> 00:07:45.120
Jirah Cox: Totally well this one also is a performance.

00:07:46.320 –> 00:07:53.070
Jirah Cox: Needle mover for CA shares right so continuous continuous availability, where I can, I can tell the FSB Adams.

00:07:53.520 –> 00:08:04.860
Jirah Cox: Pay do more work be more in lockstep but then, if an fs vm goes down the end user doesn’t even perceive a fail over there right it’s the data is immediately live on the next MSP am so even more availability.

00:08:05.970 –> 00:08:07.080
Jirah Cox: And now, with more performance.

00:08:07.680 –> 00:08:07.980

00:08:09.120 –> 00:08:13.680
Andy Whiteside: i’m never asked this question, maybe one maybe you get it all the time, maybe you don’t but.

00:08:16.050 –> 00:08:28.440
Andy Whiteside: Maybe i’m just an idiot here, so this is a mf s as in network file system or This is like SMB windows files or all the above.

00:08:28.680 –> 00:08:31.620
Jirah Cox: file supports both okay yeah it.

00:08:32.160 –> 00:08:33.960
Jirah Cox: doesn’t manifest or also you know.

00:08:34.350 –> 00:08:39.180
Jirah Cox: A dual presentation right, I want, I want to SMB nfl for the same data set.

00:08:40.980 –> 00:08:45.060
Andy Whiteside: You know my role in this podcast is do the introductions and ask the dumb questions.

00:08:45.300 –> 00:08:47.280
Jirah Cox: No good question so.

00:08:47.340 –> 00:08:53.430
Jirah Cox: it’s all about the the use case right so SMB very common in the windows world right a lot of shared drives now drive profile drives.

00:08:54.690 –> 00:08:59.370
Jirah Cox: And fs in general, more common in the Linux world application storage and data logging.

00:09:00.450 –> 00:09:05.910
Jirah Cox: All that kind of stuff but you know for some customers for some environments, you need both right, I need to write it from one system read it from the other.

00:09:06.870 –> 00:09:17.280
Jirah Cox: Maybe you know, an application that logs it under one Protocol, but then my analytics picks it up over the other protocol, and I want to do an analysis on it or.

00:09:18.540 –> 00:09:23.400
Jirah Cox: Big data crunching so that can be a use case for the dual presentation, as well the Multi multi protocol.

00:09:24.390 –> 00:09:27.210
Andy Whiteside: Alright Harvey fix your credit SMB stands for.

00:09:31.980 –> 00:09:33.000
Jirah Cox: server message block.

00:09:37.170 –> 00:09:39.360
Andy Whiteside: I don’t even think it’s I think it’s small message bug in it.

00:09:40.560 –> 00:09:42.990
Jirah Cox: This is what you will tune in for his diary googling things.

00:09:47.880 –> 00:09:52.800
Harvey Green: wow yeah you definitely caught me today.

00:09:55.410 –> 00:09:58.770
Andy Whiteside: You get it small Mrs block right that’s That was the Microsoft secret.

00:09:58.830 –> 00:10:01.830
Harvey Green: sauce I see server message block.

00:10:01.950 –> 00:10:03.000
Andy Whiteside: Oh no.

00:10:04.140 –> 00:10:05.100
Andy Whiteside: It must have changed it.

00:10:07.350 –> 00:10:11.160
Jirah Cox: Please hold while I wait for all the small to medium business results right.

00:10:13.980 –> 00:10:18.840
Harvey Green: yeah i’m seeing server manage message Bot.

00:10:19.080 –> 00:10:25.680
Andy Whiteside: See always there we go see learn something I thought I was gonna trick somebody I end up learning something.

00:10:25.950 –> 00:10:29.850
Harvey Green: yeah sorry server message box, yes, yes.

00:10:30.300 –> 00:10:42.600
Andy Whiteside: And I think the Microsoft implementation so server message block and then the Microsoft implementation of it is cysts yeah common information common something Internet file system, yes.

00:10:44.040 –> 00:10:45.450
Jirah Cox: got me there was an ultrasound that one.

00:10:46.410 –> 00:10:59.850
Andy Whiteside: Think that’s what it’s called and then you have to Google that one all right, well, and so I thought about that, because the actual paragraph ends with continuous availability and they came up with an acronym for that, too, I mean it’s is that California, is what is that.

00:10:59.910 –> 00:11:04.440
Jirah Cox: Non compliant right it’s a two letter acronym compared to ours are typical tech three letter acronyms.

00:11:04.800 –> 00:11:06.390
Andy Whiteside: Do we have to give everything an acronym.

00:11:08.220 –> 00:11:13.590
Jirah Cox: yeah I mean you know you’re thinking about your your fingers right you got a limited number of keystrokes in your lifetime right, so you got to spend them wisely.

00:11:14.940 –> 00:11:15.480
Andy Whiteside: yeah.

00:11:17.070 –> 00:11:28.290
Andy Whiteside: The problem with acronyms everywhere is then other people will use them and they don’t have any idea what they mean and then the next thing you know we just live in a world where it’s acronym and we just talk as if we’re texting each other over the.

00:11:28.590 –> 00:11:31.470
Jirah Cox: next thing you know you hear about smallest as blocks and we’re just lost.

00:11:33.090 –> 00:11:42.600
Andy Whiteside: All right next topic is cpu scheduling with nfl which stands for network file system right, yes, all right yep.

00:11:43.740 –> 00:11:45.750
Andy Whiteside: Okay pop quiz what is SA Yun standpoint.

00:11:46.620 –> 00:11:48.120
Harvey Green: storage area network.

00:11:48.150 –> 00:11:48.480
Jirah Cox: yeah.

00:11:48.810 –> 00:12:01.380
Andy Whiteside: Do you network i’ll never forget the first time that came up, I went running until my boss, how we need to buy a sand, he said, great I can afford the I can afford to say and who’s gonna pay for the network i’m like what network oh now I get it.

00:12:05.610 –> 00:12:07.230
Andy Whiteside: which went back to the IBM shark.

00:12:11.880 –> 00:12:12.450
Harvey Green: shark.

00:12:13.920 –> 00:12:19.080
Andy Whiteside: There was some IBM proprietary thing that are unix it unix team love love.

00:12:20.760 –> 00:12:21.270
Harvey Green: wow.

00:12:22.710 –> 00:12:31.260
Harvey Green: I don’t know you you’re making me feel like either i’m off or I don’t know what’s going on today, maybe I need some caffeine, for once.

00:12:31.770 –> 00:12:39.120
Andy Whiteside: If you’ve ever worked in a big time unix IBM shop, you would have no reason to know what these aren’t even at the time they were probably 15 years old antiquated.

00:12:42.720 –> 00:12:44.760
Andy Whiteside: cpu scheduling However, you want to take this one.

00:12:45.480 –> 00:12:55.200
Harvey Green: Sure, so this one is around improvements that they made to how the cpus are actually utilized when you’re using the nfl protocol.

00:12:56.250 –> 00:12:58.050
Harvey Green: You know lots of lots of.

00:12:59.310 –> 00:13:11.310
Harvey Green: Lots of I OPS back and forth for nfl was leaving it leading to high cpu usage they’ve done some improvements around the way that.

00:13:12.210 –> 00:13:30.750
Harvey Green: It limits the locking and scalable contention around the processors, just to help out with the the utilization there and, of course, you know for for everybody listening, will you processes are happy and not overwork things happen faster in your environment so.

00:13:31.770 –> 00:13:49.260
Harvey Green: Getting cpu usage down is definitely something that is a big help for those using an fs and just being able to have more more intensive workloads that you can take care of and actually just have less cpu usage overall.

00:13:50.640 –> 00:14:00.150
Harvey Green: That also helps with being able to get some faster throughput through there, so you can have some some additional ios that are happening, depending on what your workload is.

00:14:01.830 –> 00:14:05.730
Andy Whiteside: So, should this wouldn’t be cpu or bcp where’s this optimization happening.

00:14:06.060 –> 00:14:15.720
Harvey Green: So this optimization is happening that so I guess that’s that’s a great question so if I back up a layer here when we’re talking about.

00:14:16.200 –> 00:14:34.020
Harvey Green: Improvements that were made the files just files is is hosted by multiple virtual machines and i’m sure yes we’re talking about virtual machines, but at the same time that that does back into cpu the hardware cpu as well, so.

00:14:35.070 –> 00:14:49.170
Harvey Green: I don’t think is just referring to just the virtual cpu I think that it’s referring to the the hardware cpu system clock as well, I guess, let me toss the gyro in case i’m saying the wrong thing.

00:14:49.710 –> 00:14:51.420
Jirah Cox: No, no you’re spot on so if the.

00:14:51.450 –> 00:14:54.990
Jirah Cox: files as an application that you can enable within the mechanics platform.

00:14:55.350 –> 00:15:05.250
Jirah Cox: runs as a virtualized application within vm that you can just see in prism are in the Center and those Games, of course, are multiple V cpu based on your sizing and throughput.

00:15:06.420 –> 00:15:09.450
Jirah Cox: Which means that files is there for a you know, a multi cpu.

00:15:10.500 –> 00:15:19.320
Jirah Cox: distributed APP that runs within within those fs vm so this refers to my readers efficiency on the code that runs within those via FSB EMS.

00:15:20.010 –> 00:15:38.850
Jirah Cox: letting it, you know you do more work per tick right per cycle so for without growing sizes of his films in this case, one test showing a double the number of fs based I OPS so more throughput more work done without needing to allocate more V cpus to those virtual file servers.

00:15:41.370 –> 00:15:48.570
Andy Whiteside: I was having a conversation with a guy or a day the CIO at a 30,000 employee company and.

00:15:48.990 –> 00:15:56.370
Andy Whiteside: They decided hey this vdi things, what we need to do, and they went to their integrator overseas they’re managed service provider.

00:15:56.940 –> 00:16:06.210
Andy Whiteside: And said, we want to you know triple or vdi users and get really powerful virtual desktops and they were shocked at what it was going to cost.

00:16:06.810 –> 00:16:14.100
Andy Whiteside: And they really hadn’t built that into their initial offering and thought that it was just gonna be included as part of the initial offering.

00:16:14.670 –> 00:16:25.110
Andy Whiteside: And what the managed service provider that hosts their workloads told them, they would get incremental upgrades will incremental upgrades means we got we’re going to patch not we’re going to make it better.

00:16:26.700 –> 00:16:36.270
Andy Whiteside: And you know things like what you guys are doing here is going to help compensate for the lack of newer faster hardware you’re smarter.

00:16:36.900 –> 00:16:51.750
Andy Whiteside: Through software and actually did ask him in the conversation he has new tannic software hardware and after we talked through a little bit obviously it’s software and you know software just continuously making that underlying platform, whatever it is smarter asking yes.

00:16:52.800 –> 00:16:54.360
Harvey Green: continuous improvement.

00:16:57.270 –> 00:17:02.490
Andy Whiteside: Like Harvey has a tesla and it gets smarter every time he gets it update and hopefully more secure thing.

00:17:02.520 –> 00:17:02.940

00:17:06.900 –> 00:17:11.130
Andy Whiteside: Just need new tires occasion there’s a physical layer that can’t be overlooked.

00:17:11.670 –> 00:17:13.650
Harvey Green: That is 100% true.

00:17:15.240 –> 00:17:19.200
Jirah Cox: yeah, let us know, let us know when you should be like software defined tires that’d be pretty cool.

00:17:20.190 –> 00:17:23.940
Harvey Green: I think software defined tires would be known as floating.

00:17:26.400 –> 00:17:27.000
Jirah Cox: sounds awesome.

00:17:27.810 –> 00:17:28.440

00:17:29.880 –> 00:17:33.870
Andy Whiteside: All right next one here is flow control with new tonics volumes.

00:17:34.800 –> 00:17:42.360
Jirah Cox: So this describes enhancement about how again those front end the virtual file system virtual filers virtual file servers.

00:17:43.290 –> 00:17:57.870
Jirah Cox: send data back into the cluster itself, so the back end data storage new tanks volumes, which is our eyes, because he block based storage and all of the file servers in the front mount all of the various volumes in the back, and so, with adding.

00:17:59.220 –> 00:18:07.440
Jirah Cox: flow control to that working on some of the ice cozy queuing it lets the this new version files photo a little more burst the workloads.

00:18:08.190 –> 00:18:21.030
Jirah Cox: And it calls that large storage cues as well, so we can more rapidly ingest data on the front end get it back to the back end and get that stored even more quickly so good to good to streamline that and the post and run faster.

00:18:23.220 –> 00:18:24.300
Andy Whiteside: or any thoughts on this one.

00:18:26.940 –> 00:18:28.740
Harvey Green: Oh, I mean I think jarrod hit it.

00:18:29.760 –> 00:18:30.840
Harvey Green: As well as he can hear.

00:18:31.110 –> 00:18:33.570
Jirah Cox: we’re all relevant probably what fans of faster.

00:18:34.080 –> 00:18:35.580
Harvey Green: I yes for sure.

00:18:37.110 –> 00:18:46.020
Jirah Cox: And the other cool thing to call out all of these enhancements right are the things you just get with a one click software update right, so if you’re running old version of files, you know.

00:18:46.500 –> 00:19:00.750
Jirah Cox: Telling the Technics specifically lstm right just scan for what else is out there it’s new it can say hey you’re on files version X files version, why is available click here to upgrade and upgrades all automated for you all through the lcs logic and automation engine.

00:19:02.370 –> 00:19:03.930
Jirah Cox: So you just get all this goodness.

00:19:05.220 –> 00:19:06.090
Jirah Cox: That we’re talking about here.

00:19:06.900 –> 00:19:16.260
Andy Whiteside: i’ve never really thought about this, but based on this good new tannic says, I scuzzy implementation be the third tier in a three tiered storage solution.

00:19:18.330 –> 00:19:27.990
Jirah Cox: As a protocol sure it’s I scuzzy right you’ve heard about it all day every day elsewhere, the it’s the same I scuzzy that we know and love question mark don’t love.

00:19:29.040 –> 00:19:30.300
Andy Whiteside: What the acronym stands for.

00:19:30.660 –> 00:19:35.910
Jirah Cox: In this ice cozy the whoo the Internet.

00:19:37.620 –> 00:19:38.400
Jirah Cox: What I scuzzy.

00:19:39.630 –> 00:19:41.520
Jirah Cox: This is what people to inform listen to our.

00:19:41.700 –> 00:19:43.230
Andy Whiteside: story what is scuzzy I love that.

00:19:43.260 –> 00:19:46.470
Andy Whiteside: What is ngozi cuz right we have SAS and scuzzy drives.

00:19:46.950 –> 00:19:50.160
Jirah Cox: Soft know serial connected.

00:19:52.620 –> 00:19:54.330
Harvey Green: Connected serial interface.

00:19:56.700 –> 00:20:03.240
Harvey Green: I go I remember these off the top of my head anymore, I guess, because they’re not I don’t talk about them as much.

00:20:04.320 –> 00:20:06.990
Jirah Cox: Smaller theater system interface oh interesting about.

00:20:07.470 –> 00:20:10.170
Harvey Green: Oh yeah I wouldn’t have gotten there today.

00:20:11.790 –> 00:20:14.910
Andy Whiteside: server and you have to decide, are you going to put scuzzy or SAS if you put.

00:20:15.000 –> 00:20:15.630
Harvey Green: A seed yes.

00:20:16.080 –> 00:20:17.040
Harvey Green: I do remember that.

00:20:18.570 –> 00:20:25.680
Andy Whiteside: yeah SAS and if you put anything other than scuzzy all the other network server administrators look down on you yeah.

00:20:29.160 –> 00:20:30.210
Harvey Green: I do remember that.

00:20:30.900 –> 00:20:32.910
Andy Whiteside: Constantly $4,000 more.

00:20:33.510 –> 00:20:43.890
Jirah Cox: And I got the I write the Internet small computer systems interface, which I would say at this point has zero correlation whatsoever to any modern use a device ngozi.

00:20:43.950 –> 00:20:45.210
Jirah Cox: it’s not all systems.

00:20:45.720 –> 00:20:46.860
Jirah Cox: it’s not over the Internet.

00:20:47.910 –> 00:20:49.890
Jirah Cox: I guess it’s the interface but what’s not an interface.

00:20:49.950 –> 00:20:58.860
Andy Whiteside: Well, my Internet they mean Internet Protocol like tcp IP is IP is Internet Protocol, even though you’re not going across the Internet.

00:21:00.150 –> 00:21:05.910
Jirah Cox: I mean yeah but that would almost argue more for like calling it like Ellis ngozi like land scuzzy.

00:21:07.500 –> 00:21:18.390
Jirah Cox: i’m not really improving it here, but like even just the what a way to like carbon date, a term by itself computer system like two words, we can put together in the decade.

00:21:19.230 –> 00:21:25.680
Andy Whiteside: Maybe Steve Jobs there’s everybody’s just copying the number of everybody was copying Steve Jobs for a while I this and I that.

00:21:26.280 –> 00:21:28.560
Andy Whiteside: I almost called I INTEGRA i’m sorry.

00:21:31.140 –> 00:21:33.240
Jirah Cox: i’m curious about when I was.

00:21:33.450 –> 00:21:34.710
Jirah Cox: watching patented.

00:21:35.580 –> 00:21:36.390
Andy Whiteside: Oh it’s been a while.

00:21:36.570 –> 00:21:38.400
Jirah Cox: IBM Cisco in 1998.

00:21:38.610 –> 00:21:44.400
Jirah Cox: A draft standard in March 2000 So yes, I mean I checks out around around the I everything time frame.

00:21:44.970 –> 00:21:50.610
Andy Whiteside: I am I, my first three tier conversation was an ice cozy saying is what they were proposing to me.

00:21:51.570 –> 00:21:55.530
Jirah Cox: The iPod launched in 2001 yeah.

00:21:56.760 –> 00:21:57.000

00:21:58.500 –> 00:21:59.160
Jirah Cox: metrics now.

00:21:59.910 –> 00:22:00.930
Andy Whiteside: lost the answer to the.

00:22:00.930 –> 00:22:01.530
Jirah Cox: question here.

00:22:02.490 –> 00:22:04.800
Jirah Cox: You still your own answer.

00:22:06.030 –> 00:22:06.480
Jirah Cox: The.

00:22:07.740 –> 00:22:22.380
Jirah Cox: It, as I said, he today it’s not a vast to certify the vmware array for storage offload certified provider for V server systems, so you wouldn’t really want to use it as a non certified storage provider for part of a three tier design.

00:22:23.220 –> 00:22:28.920
Andy Whiteside: So what what is the example of a use case where you would connect to it.

00:22:30.750 –> 00:22:32.910
Andy Whiteside: i’ve got an idea in my head i’ll ask you the question.

00:22:33.630 –> 00:22:37.080
Jirah Cox: there’s an interrupt makes matrix on the tax portal for.

00:22:38.520 –> 00:22:44.160
Jirah Cox: The validated client systems, but basically you know broad swath of windows server Linux server.

00:22:45.690 –> 00:22:51.210
Jirah Cox: Of course we use ourselves for files actually even like the back end of like PC and some other systems all use it as well, so whatever we want to.

00:22:51.600 –> 00:23:01.950
Jirah Cox: Have like a separate data disk attached to a virtual machine was was great choice for that so usually will connect to in guest systems like windows and Linux rather than connect to like an extra hypervisor.

00:23:03.630 –> 00:23:06.900
Andy Whiteside: So when you’re building that exchange cluster and you want on their mind.

00:23:07.410 –> 00:23:12.060
Jirah Cox: Not exchange cluster I mean there’s still a thing right yeah once once in a blue moon you find them.

00:23:13.110 –> 00:23:24.270
Jirah Cox: You know sequel server that needs, you know 1012 discs or so for performance design also like healthy like log Dr temp Dr data drive separation that kind of stuff yeah yeah.

00:23:25.710 –> 00:23:27.570
Andy Whiteside: sounds like the funnel days.

00:23:28.530 –> 00:23:30.900
Jirah Cox: A lot of our systems actually automated for you, so if you use like.

00:23:31.470 –> 00:23:44.610
Jirah Cox: era to deploy a vm for a database on the tan X files does all this for you, by the way, like there’s there’s systems, I could set up like I don’t know hundreds of volumes based disks on a cluster and you would never even know about it yeah.

00:23:45.270 –> 00:23:47.550
Harvey Green: I don’t know jarrod that sounds a lot like automation.

00:23:48.420 –> 00:23:50.400
Jirah Cox: It sounds like good automation I might even.

00:23:52.470 –> 00:24:05.100
Andy Whiteside: do have to tell you guys if I did a San virtual sand project last week in my lab and the minute I found my IQ ends, they just showed up during scan I get so excited.

00:24:05.910 –> 00:24:06.450
Andy Whiteside: This works.

00:24:06.810 –> 00:24:11.040
Andy Whiteside: And I had to go type of noon, or something flashbacks in the old days.

00:24:11.310 –> 00:24:13.800
Jirah Cox: Again, you know limited keystrokes right you gotta spend them well.

00:24:15.480 –> 00:24:24.390
Andy Whiteside: Alright, the last year is pushing the performance envelope Harvey these graphs that you’re going to talk through here a little bit or not, the tesla graphs they don’t get too excited but.

00:24:24.420 –> 00:24:25.440
Harvey Green: They are not.

00:24:26.880 –> 00:24:31.680
Andy Whiteside: Give us an example of what kind of performance results they’re showing here with this release.

00:24:32.220 –> 00:24:47.550
Harvey Green: yeah, so this is talking about again, I guess, under the umbrella of continuous improvement in order for you to say you’re continuously improving, you have to be able to compare what you’re doing now versus what you were doing before you did it.

00:24:48.600 –> 00:25:00.570
Harvey Green: So you know they’re going through and making sure that they’re doing tests before and after any of these upgrades so that, when they say improvement it’s actually no proof.

00:25:02.220 –> 00:25:22.020
Harvey Green: You know i’m sure all of us have been through some you know some updates in our lives, that was not actually an improvement so they’re just trying to test to make sure that they are actually making improvements and so they’ve got some graphs on here to show the different ways that.

00:25:23.400 –> 00:25:38.250
Harvey Green: They were checking through storage these tests that says will run through FAO to determine maximum throughput for each of the configurations of one node four nodes eight nodes 12 notes and 16 nodes.

00:25:39.180 –> 00:25:45.960
Harvey Green: And they wanted to make sure that they see the capabilities of the cluster scaling up with the number of.

00:25:47.010 –> 00:26:07.530
Harvey Green: files nodes and the number of hardware knows that they have, and so you can kind of see as they go through random read random right so sequential read and sequential right that all of the graphs are going up into the right which is you know exactly what they want to see, most of them.

00:26:08.580 –> 00:26:24.780
Harvey Green: are almost completely linear so also again what they want to see as you add more and more nodes to it, you should see kind of a linear uptick in in their their ability to perform.

00:26:26.040 –> 00:26:29.550
Harvey Green: So that’s part of being able to scale up and have.

00:26:30.630 –> 00:26:38.640
Harvey Green: Some some reasonably X some reasonably good expectations, yes Andy I see your hand is up.

00:26:38.910 –> 00:26:45.060
Andy Whiteside: We did you just you said scale up, and I think what you really meant because hyper converge is all about it scale out.

00:26:46.770 –> 00:27:06.510
Harvey Green: Well, yes, so you you’ve got yes you’re correct so i’m looking through here, and I mean there they specifically mentioned that that they wanted to see the capability scale up and then that baited me in the same scale up for you to bring that up but yes you’re 100% correct.

00:27:07.560 –> 00:27:07.890
Harvey Green: me.

00:27:08.340 –> 00:27:19.770
Andy Whiteside: Is it fair to say that you’re whether you’re scaling up or scaling out, whichever way you prefer your your benefit is going to still apply and probably just the way in this linear fashion that you’re seeing here.

00:27:20.400 –> 00:27:30.480
Harvey Green: yeah so that’s I mean that’s the goal right they want to be able to prove that out, to be able to give you some realistic expectations of what things will look like.

00:27:31.050 –> 00:27:46.410
Harvey Green: As you add more hardware, as you add more files nodes so that you can have an expanded predictability for being able to get performance out of out of your new tannic files deployment.

00:27:47.880 –> 00:27:56.310
Jirah Cox: yeah it’s pretty pretty well understood commonly that you know newt annex being a scale up platform, as you add more nodes Of course you get more capacity.

00:27:56.790 –> 00:28:04.800
Jirah Cox: In the cluster but also you get more throughput right more you know we bring we bring another storage controller, to the party, it can do more io and everything just runs faster.

00:28:05.370 –> 00:28:10.110
Jirah Cox: So this is an important validation I love that we are doing right with our release testing is validating the.

00:28:10.590 –> 00:28:21.150
Jirah Cox: APP as well that runs on the platform this case files also scales out and also you know brings more performance to the party when you scale it out that way, so good linearity there.

00:28:22.860 –> 00:28:31.500
Jirah Cox: You know, in a good reminder right that you know if you’ve if you’ve started with a 10 X and you had 1000 vdi users and you double that the 2000 users and everything’s running great.

00:28:31.980 –> 00:28:41.850
Jirah Cox: Fantastic you know it could be worth looking at hey do you also want to add some more virtual file servers as well, because you can also just one click expand those as well and you’ll bring more files throughput.

00:28:42.870 –> 00:28:45.450
Jirah Cox: To the end user, environment, on top of all that.

00:28:46.200 –> 00:28:52.860
Andy Whiteside: And this is important because, going back to make it fun for our three tier friends, then this that logic doesn’t necessarily apply there right.

00:28:53.460 –> 00:28:55.560
Jirah Cox: Totally right there’s not.

00:28:56.820 –> 00:29:04.230
Jirah Cox: Probably a really a great way to start with fewer users and like less hardware and then, as you add more users, like, I guess, add more hardware.

00:29:04.950 –> 00:29:16.980
Jirah Cox: It often doesn’t scale out quite as linear when you release your left buying sort of like way more harder than you need up front and using less of it or helping you grow into it and hoping you grow past it and exceed it yeah.

00:29:17.130 –> 00:29:26.340
Harvey Green: From from a three tier standpoint that that logic is still great until about six to nine months after you bought it when they come out with a new lead.

00:29:28.590 –> 00:29:30.660
Harvey Green: Then it’s not so easy to just add.

00:29:31.950 –> 00:29:36.360
Andy Whiteside: And you may or may not get the software benefits of the new one, or the hardware capacity.

00:29:36.600 –> 00:29:37.950
Correct yeah.

00:29:40.110 –> 00:29:44.580
Harvey Green: Especially now, when you order something and it might not show up for 60 minutes.

00:29:45.990 –> 00:29:47.220
Andy Whiteside: A day before even implemented.

00:29:47.550 –> 00:29:48.000

00:29:50.220 –> 00:29:52.770
Andy Whiteside: Because that’s the end of the blog here um.

00:29:53.310 –> 00:30:02.760
Andy Whiteside: I think this just further validates the new tactics is a software company that’s helping you get the most out of the hardware, that you have in your buying in your scaling and, as you.

00:30:03.030 –> 00:30:23.880
Andy Whiteside: continue to take advantage of moore’s law, you get continued benefits of smart software and, in this case files, specific and in our world, whether it’s you know APP application data, whether it’s a user profiles, whether it’s you know the the the volume containers for.

00:30:25.620 –> 00:30:33.960
Andy Whiteside: abbeydale local or or what have you whatever is going in that fs logics partition we had a meeting last week where somebody.

00:30:34.950 –> 00:30:48.390
Andy Whiteside: is moving from one system to the other, and in that case they highlighted that it was fs logics that enabled them to get much better results, and like well that wasn’t the fault of the previous software you’re using and you just need to use fs logics all along.

00:30:49.440 –> 00:30:56.310
Andy Whiteside: But to them, you know the result was they plugged in a new smarter software solution and it all worked better, therefore, it must be better.

00:30:58.770 –> 00:31:02.250
Andy Whiteside: And that just goes along with the idea that you know the the underlying file system.

00:31:03.450 –> 00:31:15.540
Andy Whiteside: Is plays a big role I use the example all the time, have you guys both remember when you went from your first spinning hard drive in a desktop or laptop to an ssd drives you thought it was the best system ever it was just more.

00:31:16.020 –> 00:31:17.010
Jirah Cox: I would just sit there and reboot.

00:31:17.070 –> 00:31:18.030
Jirah Cox: Over and over again, like.

00:31:18.150 –> 00:31:19.080
Jirah Cox: I said again what’s this.

00:31:21.360 –> 00:31:23.940
Andy Whiteside: I did that, with a hypervisor run on envy me.

00:31:24.870 –> 00:31:30.600
Andy Whiteside: efforts on and it booted and four seconds tonight I showed my kids I didn’t get it, and I was like this is amazing.

00:31:31.740 –> 00:31:33.060
Harvey Green: yeah they won’t get it.

00:31:36.090 –> 00:31:39.750
Andy Whiteside: Alright guys well, I appreciate you jumping on on a Monday afternoon i’ll let you go back to.

00:31:41.580 –> 00:31:45.420
Andy Whiteside: Hopefully wrapping up the day and i’ll go go pick up my family from animal kingdom.

00:31:46.950 –> 00:31:48.330
Harvey Green: You have followed that Sir.

00:31:48.480 –> 00:31:49.230
Jirah Cox: yeah you enjoy.