Grey Beards on Systems - 112: GreyBeards annual year end wrap-up with Keith & Matt

Episode Date: January 1, 2021

It’s the end of the year, so time for our regular year end wrap up discussion with the GreyBeards. 2020 has been an interesting year to say the least. It started out just fine, then COVID19 showed up and threw a wrench in everyone’s plans and as the year closes, we were just starting to … Continue reading "112: GreyBeards annual year end wrap-up with Keith & Matt"

Hey everybody, Ray Lucchese here with Keith Townsend and Matt Lieb. This Great Red Dense Storage episode was recorded December 22nd, 2020. This is our annual year-end podcast where we discuss the year's technology trends and what to look forward for the next year. Keith, what would you like to talk about today? You know what? Let's talk about the biggest driver of technology, I think, for the year, which has been COVID-19 and the impacts on the industry. Yeah, yeah. This whole work-from-home transition kind of hit the world big time in a very short period of time. So what do you think are the big aspects? Certainly we've seen a massive growth in options

in terms of VDI solutions. Little companies that are putting out solid products like Dijon and, you know, I could go on and on. Citrix has matured massively in their solution. Of course, we've got VMware and NetApp has a whole ecosystem built around VMware as well. Anybody seen anything otherwise interesting in the VDI space? I think I've been sort of touting this remote work from home for about two decades. I thought I would, you know, within the last 15 years ago, I thought we would see the end of office work and stuff like that. But in reality, it's persisted throughout the decade. I think it will come back to some extent,

but it won't come back nearly as much as much as it uh once was i'm sure yeah and i think it's beyond just vdi or even office work if you look at the macro impacts you have uh i think i was talking to one of uh our industry peers peers who's in New York. He was saying for the first time ever, he got offered a rent special in Midtown Manhattan. And this is for an apartment building he's already living in. So this wasn't to attract new renters but this was to attract uh re-ups yeah so the and it's not just vdi that's kind of empowering the solution i think vdi is a great starting point but it's advanced solutions like citrus workspace and

vmware's workspace one refactoring applications to be web-first, mobile apps, et cetera. Well, the cloud, too. I mean, to a large extent, the cloud has been sitting in the sidelines. Well, sidelines is probably not the right answer, but it's been out there for decades now, and it's really starting to take off. You know, a great indicator is that, I saw a headline the other day that Amazon is a

$10 billion storage company. So how much of that is driven by demand for as a result of COVID-19 and the ability to transform? I think Microsoft has said that they had customers do three years of transformation, technology transformation in three months. And it's not just working from home. It's also changing, shifting of business models. I mean, we consume and do stuff. We interact with our vendors much differently than we have in the past. Well, the whole Zoom thing, you know, purchases, electronic online purchasing and stuff like

that. Yeah, exactly. I was wondering if anybody still has a fax machine. I do, but it's disconnected. How's that? No, seriously. But it brings in other aspects. And I think it's really important to understand that with all of these additional attack vectors,

we find ourselves in a place where security has to become a higher concern even than it was before. Yeah, yeah. Well, I mean, security exposure becomes a little bit, it certainly becomes larger as more and more people work from home, more and more corporate data is sitting on home computers than ever before. So there's that whole aspect that COVID has sort of accelerated. I think we'll talk more about security later in the show as well. But yeah, you know, so I've been working from home for almost 20, not quite 20 years, but for quite a while. I'm sure, Keith, you've been working from home

for a long time as well. It takes a different mindset to some extent to work from home than it does to work from the office. I think there's a whole psychological perspective there. I always felt I was a little bit, it was a little bit lonelier existence working from home versus, you know, you could have lunch with your buds and stuff like that because they were all in the office and stuff.

And on the other side of the equation, though, don't you feel you can get more accomplished than sitting in an office with a lot of shoulder taps and a lot of, oh, by the ways. You know, when I was in the office, we would spend like probably, oh God, 50 or more percent of the day in meetings, just talking to people about what's going on and trying to keep everybody up to date and all that stuff. I find working from home, the meetings are much more, much more select and much more focused to some extent. I don't know. I mean, what do you guys think? Nah, I think collectively I work more working from home than I do in the office.

There's a lot less water cooler time. There's a lot less idle time. Sure, I don't have as many in-person meetings, but I'm kind of strange. I didn't really mind in-person meetings that much. I like people. uh unlike me he doesn't have his his uh girlfriend works uh retail so she's not there during the day so he is extremely depressed and and he's not suffering from depression but he's depressed you know he's really bored his work day is mundane so it it has impacted his productivity some as well. So I think it's a mixed bag. I think there's definitely not a one size fit all solution for companies who will have to figure out a balance coming out of this. But there is definitely for some people is great for other people. Not so much yeah i would think the guys like from chicago and stuff that were commuting downtown

or worst case might be san francisco commuting to the valley or something like that that sort of thing was was uh insanity right yeah so you get a you get an hour or so each way back in your day but don't you find that you're actually working those hours? It's not like it's a gift. If anything, it's more like a gift to the company. Yeah, I don't know if I'd say it's a gift. You know, of course, I'm my own company, so it doesn't matter if I give stuff to the company. It's all for my benefit. But I think, I don't know, you know, I start doing more structured things in my day. You know, I start, you know, I start spending more time exercising in the morning before work.

I spend more time, you know, just out in nature and stuff like that, just to try to get some psychological benefit out of it. You know, I don't know. It's just, I sort of structured my day. You know, in the old days when I was in the office, it was a very structured day, but it was meeting, meeting, meeting, meeting, no meeting, meeting, meeting, meeting, meeting, no meeting kind of thing. So it was that structure versus, yeah, I work out in the morning, I have breakfast, stuff like that. I don't know. It's a different... I don't think I'm working any longer hours. I'm just being more productive with the hours that I'm working. I would say, in my case, it's both. I'm surely working longer hours. But in addition to that, the hours that I'm working are far more productive. And I own my own business. So Ray, you have a very good point that

it's for my own benefit for the most part. Whereas if I'm ever an employee and if you ever consider hiring me, I don't believe in returning more value than you provide from a monetary perspective. And I expect that of the people that I hire. The work is an even trade of value. And if one person should not be over, should not win in that value transaction than the other. So if I get paid to deliver a product or service, I deliver that product and service. If I work another if I work another hour or two or 12 while working for myself, then, you know, that's a different value prop. Yeah. Yeah. Yeah. I think the other thing is that the whole office architecture of the cities, you know, cities are, you know, to a large extent, they're, they're major office centers,

they're major apartment centers, and they're major retail centers to some extent to satisfy the apartment dwellers and the office dwellers. But the whole office world is going to be, I think, is going to change over time here. It seems to me that, you know, that's got to shrink. Well, in what way? You think the footprints of the offices are going to shrink? Yeah, I think the number of offices will shrink. The number of the need for office space will shrink. Whether you go to this, you know, I don't know what they call it, hot desking or whatever,

whether you share desks for employees or not. I always thought that you really don't need in this IT space as much on-premise people to do the work that needs to be done. Yeah, there needs to be some, but yeah, I don't know. Keith, you have a lab. Do you have people there all the time? No. I have a data center engineer that comes up probably once a quarter. I only go into the office a couple of times a month. I actually have a office space down there, shared office space down there.

I only go in a couple of months. My team is completely virtual. So, yeah, there's not a whole lot of need. I will say there is something to be said about having a permanent space, whether that's in your home office or in the office. I did the whole hot office share thing when I was at PwC because we were only in the office one day a week at best. And it's not a great experience. Like it did not encourage me to go into the office. Matter of fact, I avoided the office because I don't know. I didn't ever know if I was going to be able to get a desk or not. Hmm. Hmm. Yeah.

So you think a reservation program would help that or something like that? Yeah. And it doesn't. People don't respect it. I gotcha. It's kind of like, you know, the people who work in the because you still have some people who work there like Monday through Friday and they just, they just stay at the same desk and ignore the reservation system. So I reserve a desk going and, you know, there's a, I go to the desk that I thought I had reserved and there's pictures of somebody's kids there. And it just, people culturally, I don't know if we're ready for that culturally. Yeah. Well, I think it's coming, but I don't know.

Huh. You know, I did some lab work for a customer this year and normally it would be, you know, I would either have a lab here or I'd be renting a lab space in the area. It was completely remote, 100% remote. I mean, it was on the East Coast someplace. And I was in, you know, mountains area. And it worked like a champ. There was really no reason for me to be there. So, yeah.

Unless you need to physically fix or install equipment, there is no need to be on site. Right, right, right, right. All right. Well, I think the COVID thing is going to be with us for another year or so at least, and I'm sure we'll have more insight as time goes on. But the next thing I thought was technologically interesting was this whole and VMware kind of on their VM world talked about the DPU as being something they were going to they were going to port ESX to and and you know they were developing support for it throughout their their

their VMware solutions and stuff like that but this this whole SmartNICs, DPU, ARM thing has really emerged over the last couple of years, I would say. Very intriguing. We talked to the Mellanox folks, you and I did, Ray, on one of our podcasts. And I went back and listened to it recently again and I really find it to be incredibly intriguing what Mellanox is able to establish actually on the NIC itself and how they're isolating specific types of traffic and even more specifically types of transactions against that traffic on those NICs. Certainly, Mellanox has been a groundbreaker in terms of networking before. Surely, InfiniBand is a perfect example of that.

Yeah, and then if you look at kind of this whole system grown up is AWS Nitro. They've been doing it for a few years. So they take the same DPU across several instance types, and then they can take that DPU and have it deployed for specific things. So, you know, increasing speed of encryption, encrypting data in transit, storage services. So, you know, they might have a set of BYOD type storage arrays and they just built that functionality,

not in a hypervisor, but in the DPU and the SmartNIC itself. So, you know, it's nice to see companies like VMware taking these cloud scale approaches and making them consumable for the average enterprise where we just won't see it. I thought that we would see it in the form of seeing ESXi running on a smart NIC, and the reality was no, it's not ESXi running on a smart NIC. And the reality was, no, it's not ESXi running on a smart NIC, but it's extending the hypervisor functions out to the smart NIC. So NSX co-running on a smart NIC as opposed to having ESXi itself running and VMs running on a smart NIC.

Yeah. Yeah. I mean, it seemed like they're moving some of the security stuff out there. They're moving all the data processing stuff out there. It seemed like Pat really said he was going to port VMware's ESXi to run on the DPU, but that may be a long-term vision rather than a short-term goal, I talked to them and that's not the idea anymore. The idea isn't to have ESXi running on the smart NIC and virtualizing the network functions. So you can have something that has the power footprint of an Intel NUC, but this super deep power to process networking functions out at the 5G edge. That's the dream. And yet running ESXi on that is not kind of, they figured out that running ESXi on that is not the approach. The approach is to have like a single version of ESXi across, you know, your minimum three node cluster, and then extend that functionality to,

to agents or smart NICs downstream, which is much more appealing to telcos from a licensing perspective, because licensing ESXi and telco has been the problem for VMware. So SmartNICs is just disrupting a lot of how we've approached delivering data center type technologies, especially in the telco space. And you think it's as much a licensing workaround as anything? I mean, it's part of it. It's super powerful functionally to say that I am going to, I have my hypervisor control plane for my networking,

which is a pretty good abstraction. But remember, VMware ESXi is licensed per core count. So if I have a bunch of Xeon processors out in the telco edge and I have to license VMware via those Xeon cores, that's really expensive. VMware needed the telco space. Pat has been talking telco for the past five years and VMware has not penetrated the telco space. Pat has been talking telco for the past five years, and VMware has not penetrated the telco space. A big portion of that is cost, and other drivers such as telcos really want and prefer open source versus closed source systems. But pricing and licensing has been a very big hurdle for VMware entering the technical. Right, right, right. Well, I mean, the other thing is that the ARM processor has become much more mature

and much more competitive in the CPU space than it once was. There's a whole, you know, I wrote a blog post last week or the week before on hardware innovation speeding up. And it's, there are multiple facets of that. But one was that, you know, Moore's Law and the Denard scaling is kind of slowing down. Or Denard scaling has stopped almost, but Moore's Law is slowing down. So whereas before they could put more and more technology and more and more transistors and more and more instruction complexity into the x86, that advantage is slowly slipping away. Well, they're accommodating for that, Ray, by multiplying the number of cores. Yeah, I mean, and that's okay, but it doesn't increase the functionality or the speed of that core.

Yeah, Ray, we have that one podcast where the vendor tempted the death of x86 in general. I think we're a few decades away from the death of x86. the Dell guys. But the whole system on a chip approach, that vertical integration, what we're seeing in the client side with the Mac M1 and what they're able to do performance wise for a code optimized for ARM and that system on the chip design. I have a monster 16-inch MacBook Pro with an i9 processor, 16 gig of RAM, and the M1 for video editing in Final Cut comes close in performance to my i9 processor. Werner was on stage at AWS reInvent doing his keynote. He talked about Gaviton, which is AWS's ARM alternative, and how you're saving 40% on power, cost per watt, et cetera, for performance for optimized code however he also said that it was an investment if your company was looking to leverage arm and gabiton that it was a co-investment with aws

because non-optimized code as you can imagine is pretty slow. Premiere running and encoding something in Premiere, which is a video editing suite, I saw an example how it took an hour and 30 minutes on an M1 versus 20 minutes on a i9 or i7. So you get the complete opposite for code that's not optimized. And there's an awful lot of not optimized code. Right. That's interesting. I always thought that was a compiler function kind of thing, that you just get a compiler that would optimize for the processor and stuff. But there's more to it than that, apparently.

Apparently. But if you're programming in Go or Python and some of these higher abstracted languages, there's really not a reason not to go with uh especially if you're not talking about needing telemetry data or specialized io where you're programming the entire distributed system you're only programming at the you know for the compute layer where you know it's kind of a no-brainer if i'm going to write goal or if i'm

going to write python my python is going to compel uh just as well on arm as it would on uh on x86 yeah yeah that's interesting interesting yeah and so even arm's gotten more multiple cores per per system and stuff like that it's uh it's pretty impressive what they can do with ARM. And I think to some extent the smart NICs and the DPUs are also taking advantage of the ARM processors to put generic computation out there. You could run sort of container workloads on there almost. The other guys that were doing that sort of thing

was this computational storage. And we talked to them, oh God, a year or two back. But even the Flash Memory Summit this year, they started talking some more about how computational storage is taking off. But in my view, and I'm not sure whether this is the case or not, is they probably have an ARM processor sitting on the SSD doing the compression and deduplication and encryption of the data. Unique tasks that it serves better. The traditionally GPU-type functions. Right, right.

It's a similar concept to GPU where you have an acceleration, and the GPU sense has got single instruction set, but multiple data streams, whereas in the smart NICs or computational storage, they've got special purpose hardware sitting there doing things like for the smart NICs, I assume it's, you know, it's, it's, it's data transfer, byte by byte, encoding and decoding and deframing and reframing. And I think a lot of it has the more like the deduplication and compression factors become a part of that, particularly when that NIC is dedicated towards storage or WAN infrastructure. Deduplication becomes a really big function. and F5 and companies like this start to incorporate these kinds of algorithms into... For their WAN optimization?

No, the network functionality. Oh, I see. Yeah, so we were already seeing that. So, again, AWS is doing a lot of this. You know, you're talking about WAN acceleration, compression, security, protocol optimization. So if we go one step deeper into WAN optimization. If I can process a packet and if I can make a decision on a packet's direction at the SmartNIC instead of sending that to the Xeon core, even if I'm taking advantage of DPDK, I'm going to see advantages in speed. One of the things I like to look at is video encoding. If I can, if I can, if I make the request for the bits off of the disk and I can,

and at the processor, I can recognize, oh, I know that that's video. The CPU got a request to re-encode this video for an iPhone screen instead of an iPad screen. Instead of sending that workload to the CPU to do the work, I'm doing the work at the SmartNIC. That is an incredible amount of optimization and efficiency from a Wall Street right. And then, Ray, you know, we talked about computational storage. What happens if that request never even hits the node if it's done at the storage? If the SSD has this cheap to free ARM core that can do this work at the SSD, why not? Yeah, yeah, exactly, exactly. Jim, Jim Handy, which was on our Flash Memory Summit wrap up talked about, you know, it's, it's a, it's a computational intensive activity. If you can do that all outboard, why not? I mean,

obviously it's the cost of doing it, but you know, there's an advantage speed wise to doing it as well. It's amazing what, what they can do with ARM these days. Yeah. I think it's just amazing to challenge. You know, we talked about this in the pre-show a little bit, that Intel AMD x86 world has never been challenged like this in any significance. Probably in just the life of x86 as the core of compute in the data center.

The only real challenge was another Intel technology, Itanium, which- Well, PowerPC would be the IBM version of- Yeah. Then some PowerPC. So there is some, there's actually a bit of AIX running on Power and actually running SAP HANA, of all things, in data centers. But again, not a serious challenge to the juggernaut that's x86. I think to some extent, Power and Solaris and those kind of guys, and even HP, had always gone down a proprietary route. Lately, IBM has opened up PowerPC to be an open power kind of solution. ARM, to some extent, anybody can license the IP in manufacturing.

I mean, it has been open since the get-go from a hardware perspective, I guess. So yeah, it's, uh, it's taken a while and it's taken sort of this more slowdown and Denard scaling, you know, ceasing to exist. And, and, you know, it's, it's, it's, what's coming back is hardware design is becoming an advantage again. It hadn't been an advantage for a long time because, you know, the CPU. You touched this on recent podcasts, right? You were talking about how software is going to consume the world.

Yeah, that was always the case, you know. Well, only only the last 10 years or so. But what really is is true is it's going to be some combination thereof. And the hardware needs to be in place to support the software to do the job and vice versa. So, yeah, I don't think we're out of the hardware game yet. I don't think we'll ever be out of the hardware. If anything, it's accelerating. I mean, you could see that in GPUs.

The NVIDIA GPUs are coming out faster and stronger and quicker. The whole AI neuromorphic or even AI just normal accelerators, the TPU and Cerebrus and the GraphCore and those kind of guys are just loading up this hardware functionality onto these chips and taking off. There's no stopping it as far as I can tell. I don't know. Obviously, it takes a certain amount of volume to make a business out of these things, right? And I'm sure that there's that challenge. And to create a processor die set,

it's no trivial feat, right? They've been doing it for decades and decades. But it seems to me that when creating something completely brand new and to create a die for that proc to actually function, aside from testing alone, you've got to create a fabrication facility. I mean, we're talking about a lot of- Yeah, but the fab facilities now are available. You can you can hire a fab. I'm sure it costs millions of dollars, but it's not like you have to buy and create a whole fab and billions of dollars. It's not a big deal.

I know, but I wrote a blog post a couple got a couple of months back about a company. It was a fab in in in Arizona, I think, that was offering free chip development. It wasn't state-of-the-art, you know, 8 nanometer technology. It was 130 nanometer technology. But if you followed their tool chain and stuff like that, they would manufacture your ASICs for you. You know, like an account of 50 or something. If you wanted more, you had to pay for it. And, and Google, Google was somehow supporting this, this technology with,

with the software tool chain and stuff. It was pretty impressive, but again, it wasn't state-of-the-art technology, but you know, you can hire global foundries or Samsung or something like that. If you've got the money, they'll do it. And it's state-of-the-art technology. It's beyond the Intel node as far as I can tell. Yeah, I don't know if they announced it or if it was handed to, but Microsoft is designing their own ARM processor. It is definitely a good business to get into.

I don't want to slight the IDMs of the world. These integrated chip manufacturers like Intel. Intel, I think the industry has gotten to the point that we're kind of slighting Intel. They've had some slip-ups going from 14 to 10 and now 10 to 7 nanometers. These guys are some of the smartest people in the world. with 10 90-minute meter processes off of AMD 7 just shows to you once they get their act together and they get to 7, we're going to have fun in processors for years to come where we have AMD, Intel, Nvidia, Apple all leapfrogging one another. This kind of reminds me a little bit of the race to gigahertz,

if you guys can remember that from the 2000s. Yeah, we're all older than you, Keith. Yeah, you're older than me, but that was a snapshot in time that we all were. Yeah, we were all active then. All right, so we kind of killed the processor off here. So the other big story of late has been a security exposure. I don't know who wants to start in that space, but there's been a serious, serious hack it's this infamous supply chain hack that we all been worried about the super micro hack where they put a chip inside of the uh baseband of the

of the board that no one could find it was probably just bad reporting uh but what actually happened was that this happened in software. If you're very reliant on getting DAT file updates, system updates, software updates from your trusted vendors, by all indications, infiltrated SolarWinds to the point that they could inject their own code in a signed Windows DLL for their Orion network management tool that's used by evidently a lot of people because the update went out to 18,000 organizations. thousand organizations uh and i think that's 425 and orion is used in 425 of the uh as in yeah yeah the uh fortune 500 uh they then used a additional attack which was based off of a trusted security token to mimic a administrator, create local accounts on systems, and basically own the network of the companies they chose, the companies and the agencies they chose to infiltrate.

Nasty hack. Yeah. So when you say supply chain for a software vendor, what do you think? I mean, you know, a lot of these guys use a lot of open source and things of that nature. But, you know, so they would might have some IP from a, from a special software vendor that does something specific to that. And they somehow got into that vendor's software environment and were able to inject their own code? Yeah, so if you think about

just the CICD process, from what I'm hearing is that somehow the intruders injected themselves into SolarWinds' CICD process and injected

code into that workflow. So the update that went out in March, uh, It was a perfectly cryptographically signed update and it was all valid. It came from SolarWinds and everything, but it happened to have some bad code in it or some hack code. The hack code and, and, you know, to add insult to injury, uh, maybe some virus scanners might have caught the malicious, malicious activity, but SolarWinds, uh, uh, implementation details ask not to, uh, not to scan that directory.

So even if the virus, even if your virus software was smart, would not see it. And then they did not actually start the. It was set up as an exclusion. I did not know that. Yeah, it was set up as a scan exclusion list. Then on top of that, the malicious activity did start into a few weeks after the update. So it would be hard to trace back, you know, like, well, where did this come from? I see SolarWinds is doing something that it normally wouldn't or shouldn't do.

It was, it's very subtle and actually extremely genius. I was, I was blown away reading the details of it on the FireEye site. Well, I mean, and you know, the challenge is, it's just like the hardware space it's, it's gotten to the point and, you know, the challenge is it's just like the hardware space. It's gotten to the point where, you know, our processes to develop these systems have gotten so automated anymore and need to be. I mean, these guys are, you know, some of these guys are rolling updates every day to their deployments. And SaaS is even worse. I mean, this is, this is on-prem software, right? But I mean, the, the, the SAS solutions, uh, updates roll out extremely frequently. Yeah. I just rolled out a new website and I, it took me, it took us three and a half months to roll out a new website. So I couldn't imagine the complexity of getting to the point where you're doing updates every day. It takes automation.

And that automation, if you're not auditing your practices and your access, et cetera, around automation, it can get scary. So this distributed, going back to the first story, COVID-19 and this forced work from home environment that we're in and the bigger attack surface. What do you guys think about the bigger attack surface that we have now that everyone is remote? So now that the corporate data is sort of proliferated out to the field, home computers and laptops and stuff like that and the cloud and stuff in order to be able to access, yeah, it's gotten to be a larger exploit surface. Yeah, exactly. Yeah. You know, before letting the kid poke around on websites on my computer was one thing. But now then them poke around on websites on my computer. And now that, you know, we talk about the advantages of VDI, remote work, BYOD. But I'm sure there's a bunch of scared CIOs are out there like, wait, hold on.

Some random Android phone or tablet is going to connect to my network and use my mission critical apps? Yeah, it's a little scary. And the truth is, it's not just maintaining a powerful firewall and controlling the traffic in and out of your actual network, it's far more robust than that. This spoofed DLL that you were just talking about is a perfect example. Well, I can't say none, but certainly we know of only one virus software that has come out and said, we could have seen that activity should it have passed through our gate. Had it not been excluded and all that stuff. Yeah. Had it not been excluded, right. But if it's every piece of data that gets transmitted

through a VDI session or a VPN session, the quality of your scanning software has to be so robust as well as so powerful so as not to drag the network or the wide area network traffic to a halt while somebody is doing these things. it really draws itself in a very strong image against what traditional VPNs or firewalls did. Perhaps the VMware approach where they effectively create a virtualized desktop on your desktop and control access and control security seems to be a good way to go in these times of security issues and having corporate data be all over the world and stuff like that. Yeah, that's a layer. And the VMware security guys hate when I talk about VM escape. But we've seen instances of VM escape now. So, especially on the desktop, what we're running on the desktop is not nearly as robust as what we're running in the data center.

What do you mean by VM escape, Keith? So, the VM escape is when you escape, when you, when, you know, so in theory, if I have a isolated OS running inside of a VM and, and abstracted hardware, virtualized hardware on my PC, in theory, I should be able to download malware and whatever, just as long as my security settings are tight on the hypervisor, I should be fine. I should be able to do all the testing that I want. A piece of malware

can't escape and infect the larger machine. Well, a couple of years ago, that happened, especially on workstation, that the malware could escape the VM and compromise the bigger machine via like a virtual floppy. Right. At the same time, Keith, it also traversed the other direction. So from the infected machine, somehow that infestation was able to traverse into the VM that happened to be a VDI instance on the corporate network, at which point you had a transaction that took place crossing the lines of physical to virtual and then somehow traversed into the corporate network. Yep. That's why now on my desk right now is a good old fashioned pencil and paper. Yeah.

I can't read anything that I've written in it, but it's a challenge. Yeah, I do that too, Keith. I'm still in that mode of operation as well. Well, this has been great, folks. Any last items you'd like to discuss before we leave? Sorry, Ray, I got nothing for you. No, I'm good. I think we beat up 2020 pretty almost as bad as 2020 beat me up

i think i think there's a lot of potential in what transpired in 2020 i obviously the security stuff is going to take some time for us to understand and and try to resolve and make make better but at least we know about it now and we can start thinking about what it's going to take to get us there. And this whole COVID thing, the digital transformation that occurred as a part of that, I think is good for society.

I just, you know, it's just a question of making sure everybody's got an equal shot at it. And the ARM thing, I think any competitiveness in the cpu space is good for all so yeah i think so all right gents that's it for now bye matt bye keith bye ray bye ray and until next time. Bye, man. Thanks, guys. Bye-bye.