Ambient Mesh, the 90% of IT and the 2% of sysadmins, and an uninformed look at GitLabs

Episode 239 · Thu, 29 Sep 2022 12:41:21 +0200 · 45:10 · SRT · archive.org · all Tanzu Talk

I come up with a new epithet that this is a evidence of that is there's a lot of things that I should be doing, but I've gotten good at not doing them. And to that end, yesterday, for some reason, I got really interested in the age-old question of how much on-premise IT there is versus how much is in public cloud, which I think I don't know about YouTube, but I feel like that is one of the major existential philosophic questions that started 10 years ago and probably will go for at least 10 more years. It's just like the answer to that question drives pretty much everything in enterprise IT at the moment. Or, well, I don't know. I'm going to overstate it and say pretty much everything. It's not actually the case. By pretty much everything, I mean, I don't know, 60%. But is that a super majority? 70%, right? I forget. Yeah, no comment. But I know what you mean. I remember being on lots of projects back in the day that were all building up to data center closed down sort of stuff. You know, I'm going to close our data center. This is it. This is the big time. We're going to get rid of our own data center and we're going to move everything to the cloud. I remember being on loads of those and didn't once experience the data center shutting down. Yeah, yeah, yeah. And then moved careers, you know, and did other things. But yeah, so what's the latest then? Well, I'll tell you, I wanted to bring this up at the beginning here to get your two's take on this. Let me summarize what I sent a rambly unedited memo on this in my midweek edition of my newsletter. But let me summarize what I found. One, first, there's a huge provisio here. I did not spend any money on industry analyst reports. And I feel like if I had a spare $2,000 to $5,000 kind of banging around in my pocket that I didn't need, I probably could have gotten the answer very quickly. But whatever.

You could have got one answer. Yes, yes. But just getting an analyst PDF or Excel spreadsheet, that feels like cheating for this question. I mean, if you could do that, then, you know, it just like wouldn't be exciting. So instead, I did search around the internet and I found the usual press releases and news summaries of analyst stuff. And I don't know. And I'm not going to go over what it is in detail. But I feel like at this point, we might be at a 30% to 40% in public cloud. And then whatever the rest of that math is on premises. Now, the issue there is then you have to make a big, either a donut or a pie chart, depending on if you like things in the middle, of like what types of workloads those are. Because I feel like the majority of that would be what we call software as a service or just applications by revenue, at least. And I think, on the other hand, if you were to count by like actual applications, like processes, especially ones that companies build and run on their own, I kind of have no idea what that number would be. I think it might even be like much, much lower. Because maybe we have bias because we're talking with a lot of people who are modernizing their applications. But I feel like it's very rare that I come across organizations who are like, oh, yeah, we're running all of our custom built applications in the public cloud. There are many, as you were saying, Ben, who are doing everything or multi-cloud, as we like to say, taking on that cloud chaos, to use that fun phrase. But, you know, it seems like it's pretty low. Now, in contrast to that, right after I decided I was spending too much time on this and I sent out my rambling missive, I went over to my other podcast Slack channel and someone had posted a mad money with Jim Cramer interview with the CEO or head or whatever he's called of AWS, where hopefully this wasn't just offhanded because I feel like it would be accurate if it was well studied

and was backed up by a bunch of internal work. But the AWS CEO said that there's only about 10% of IT in the cloud. So there's a lot of TAM out there to eat up, a lot more out there, which I was told, as I like to say nowadays, never to do math in public. But if I did the math on that, that means that 90% of IT is still on premise, if we take that on faith. So we've got anywhere from, yeah, I would say anywhere from 90% to maybe 50%, maybe 60% of IT is on premise. But you two, what are, now, Ed, you cover the industry very closely, thankfully. Someone has to, you know, amongst the three of us. Like, what do you encounter in this area when people are trying to get a hold of, like, how much public cloud versus on premise there is around? So the first thing I encounter is all the analyst stuff. So you learn very quickly, I think, to take all that with a pinch of salt because it's probably accurate given their context, whichever analyst firm it is, right? But the numbers there vary forever. And it's one of those things I think everyone tries to keep a feel on this one, right? I certainly have always tried to keep my idea of where do I think it is. I can't remember. Well, I say I remember, but I don't remember exactly, Pat Gelsinger standing up at one of our conferences and putting up some figures around how much was cloud, how much was on prem, and predicting where it would go. And it was a very, very slow sort of growth to the cloud or slow migration. And there was a huge amount of flack at the time on, you know, Twitter and everywhere else that, no, this was just VMware saying it's all slow and everything. And then I remember sort of thinking, fast forward five, six years, and actually his were about right. It was a lot slower than at the time the AWS, everyone else was saying it was going. And I wish I'd kept those actual figures so I could see, was it exactly right spot on roughly? But definitely slower than a lot of people have said. I mean, I could benchmark my stuff on my home lab. I used to have about, probably at its peak, about six servers in my home lab. And now that's gone down to two servers, really, which are both currently turned off.

And I run my unified stuff for my networking on a cloud service. And that's about it. I turn them on occasionally. I still like the idea of being out of them. The rest has gone cloud. And I just think, you know, I'm a microcosm of everybody. So those are the actual figures. Yeah, yeah. So do you, like, I'm going to throw a few numbers at you here. Do you feel like, do you feel comfortable with a 50-50 split? Or do you want more like a 10-90 split? Which would you go closer to? I'm going nearer to probably sort of 25% cloud, 75% on the road. That's good. That's good. You know, I would be comfortable with that. Without getting too nuanced about workloads and all this business, right? Because then once you get into, like, storage, you're just like, I don't know. This is meaningless, right? Like, just because, like, there's so many different uses of storage and things like that that it becomes, like, I don't know. And it becomes odd. Now, you gave a little bit of, I think, your thinking here, Ben. Like, over the years, you've done this. But also, you know, you've been involved in, before your current role, and also in your current role, you've been involved in a lot of, like, putting your hands on keyboards and mice and, like, coding up this stuff and moving things around. So you've got a good career of experience of, like, we sure would like to run this in the public cloud. Let's see what happens. Like, what's your sense for, maybe, I don't know, you can erase this framing of the question. But when it comes to, like, the more widely used application architectures, are they, like, ready to all run in the public cloud? Or are they still sort of, like, better not making that decision at this point? Which is not to say they would run on premise. It's just, like, it could be cool either way. And you've got to figure out what is best for the application. I think, yeah, I think that's the crux of the issue is there's so many angles to it. So I think the first thing I would say about this, is it 10% or 50% in the cloud, is that

that sort of assumes that, the question, if you like, sort of assumes that there's a fixed volume. So there's a fixed amount. And, like, okay, 10% of it's over here, 9% is over here. That's not really how it is at all. You know, actually, there's been a huge amount of growth in various software and services, especially over the last 10 years. You know, we all talked about smartphones and the, you know, accessibility that that brings to applications, say, for your bank or for your, you know, your email, all sorts of stuff, everything, right? It's all in your phone. They were all additions to what was already there. So that was in addition to stuff that was already built. So every year, there's more and more stuff getting built. But it's in addition to stuff that they already have. So I don't think the number, the volume is fixed. I think that 10% is possibly true, but yeah, it's hard. And when you see it inside organizations and they're trying to move forward, it's very difficult for them to do that because they have such a complex landscape inside their organization. I like, I like, I always like a moving target confounder that, like, you know, means that you've got to, you've got to instead establish like an ongoing model that you can update that takes in the, the new inputs and is updatable. Yeah, exactly. I think, yeah, it's, it's, I would say it's probably, yeah, maybe, maybe sort of 20, 25%, like Ed said, but also the industry itself is also changing. There's been a lot more talk recently about things like sovereign cloud where, right, actually, you know, public isn't right. Private data center isn't scalable. So therefore, you know, we need something that's kind of in between something that a government

or a particular territory or region or whatever can sign up to as being okay, that's, that's what we see as being the right answer to this. And that's also, you know, muddying the water somewhat because it's not sort of, is it public? Is it, is it private? It's sort of somewhere in between. And you might never know what's on there because, yeah, I mean, I mean, to, to do the annoying thing of like going back to the fundamentals, like, I think, uh, you know, to, to the, to the like strategic choice you make, I, there's two things that you're making me think. I mean, one, there's, well, there's three, there's always cost, which is just like, yep, cost is cost. Like, you got to figure that out, right? Like there's no, I don't know. That's generally the most straightforward of things. But then two, like there, there's the, uh, the, the, the, the using whichever infrastructure stack has the capabilities that are best for you. Right. And, and that reminds me of like several weeks ago, I was talking with someone that was like in it at one of the, uh, uh, tax parts of a government, you know, the people who collect taxes, arguably every country's most favorite, uh, government ministry or agency, uh, which wherever you are, and someone else had like heard a way that they did their architecture for tax collection, which was just like, it was like genius because in order to scale up to this one-time demand for like taking on taxes, right? What they did is they realized that you can basically run the, the front end for collecting taxes, just like authenticating and uploading your tax information all in the public cloud. And then you can just batch up that data and download it to your own data center. Because really when you file your taxes, it's not real time, right? Like, it's just like some indeterminate amount of time later, we'll get back to you. And so like, it's a great use case of like, of like the, uh, scaling up nature that you have a public cloud, which, uh, which, which seemed nice, uh, you know, pretty, pretty great.

And then, you know, to, to be frank, the second point I was going to make, uh, I've completely forgotten. So maybe that's, that's a good, was it just, I think it was possibly note to self must file taxes because that's what, uh, I said, I started thinking we, we can, we can, we can, we can, we, we can talk about my exciting to-do list related to taxes in the two countries I pay taxes. And after we're done recording here, when we're off the air, that's, that's always, that's always a thrilling topic for me. Let us know in the comments, like, have you shut down a data center? Come on. You know, have you, have you done it? Have you achieved that mythical golden, uh, amazing, uh, accolade, you know, achievement unlocked data center powered down, let us know in the comments. Cause we'd love to hear about it because, uh, yeah, it seems like maybe they're not shutting down as quickly as some might think. I don't know. Yeah. Who knows? I mean, my, my desire, and you'll have to tell me, Ed, if you come across this with your well-salted analyst, uh, is, uh, like, I really want that like workload breakout, right? Cause like, it's kind of, it's kind of like Ben's moving target thing of like, I don't know. I don't really care about like, you know, applications that you buy like ERP stuff. And like, as we used to call it COTS applications, cause it's kind of sort of like, eh, unless you're like the military or spies, you should probably do that. Right. Like, it's just like, it's almost, I guess what I'm saying is if you're using like software as a service, it's almost not even interesting from this cloud question anymore because you should just do it. But when it comes to all these other workloads, uh, I don't know, that is, that is, uh, more interesting. Well, anyhow, speaking of all the other workloads, we should, uh, we should do as, as we're, we're chartered here by ourselves to do and, uh, and, and checking on the news and our, our, uh,

our funny little space here now as, uh, you know, to, to, to, to Ed is the one who gathers most of this news. So first of all, that's fantastic. We get to read all of his, uh, his, uh, scurrying about and summarizing, which is just delightful. That's right. That's right. It's, it's good stuff. And so one of the, one of the, the things that, uh, that you highlighted was this concept of an ambient mesh that I think, uh, is essentially, uh, kind of latching, not latching onto, but kind of rethinking about how service mesh is done in the Istio world. I shouldn't say rethinking. It's, uh, kind of a re, would you call it a refactoring, kind of changing it around a little bit to, to make it run. Yeah. But, but instead of me trying to summarize something that I only vaguely know about, why why don't you, uh, tell us what this ambient mesh business is? Yeah, I can summarize something I don't fully understand either. I'm okay with that. Um, so, I mean, the simplest thing for me, for my understanding, cause I'm not an in-depth network engineer is it's all about the side cars or more of the lack of side cars. So, you know, ever since they've been around service measures always needed an extra sidecar injected into alongside your pods and that meant performance impacts overheads, all the usual kind of data operations of monitoring, managing all those extra workloads that aren't really doing all that, you know, they're providing functionality on their own. And they're not a thing you would choose to have, but that's just been how it's done. Um, and now they've introduced them away in which you will still have a choice. You can still run with sidecars if you want, but you can run it without the sidecars. And some of those functionality moves to running kind of per node instead of right down in the weeds, um, which again gives you, you know, easier ops and potentially better performance. There's trade-offs for it. I think that's the point. Um, and it's, you know, there, so this is now being announced as part of, uh, of Istio, but actually about two months ago, I think I surveillance with psyllium, the drone service mesh, which has the same thing. So it's, you know, it's not just Istio, I suspect there's going to be, all of them will go this way of there is an option here. Sidecars obviously add enough overhead that people would like an option without them.

And pick and choose their trade-offs. Um, but it certainly made a lot of headlines. A lot of people talking about it and I guess it's for me, it's just that evolution. The service mesh stuff has always had a reputation for being a bit complicated. And this is one step to maybe dial it back and go for people who can accept those trade-offs. This might be a better balance, make it easier to deploy. Now, now Ben, I'm going to ask you the big hairy question here, which, which I like to ask spring people a lot is like, doesn't like spring do all of this already? Like, like, like, and, and, and that's a facetious way of putting it. But I think, I think with service mesh, the question I'm always trying to figure out is like, like there are so many ways of like doing what a service mesh does that how do they all kind of layer and fit together? Right. It's almost like, uh, I think we need to introduce like layers eight to 10 in like the network layering stack. Cause we keep like messing, we keep messing around at like layer seven, which I think is the actual application. And I think, I think what's going on in Kubernetes is like a layer 6.5 or something, but we need, we need some more. This is kind of like a Fahrenheit versus Celsius conversation where at some point you need a lot more precision, uh, even though the starting temperature is totally bonkers. Uh, but we need to figure something out there. So how, how does like, how do the concepts of service mesh kind of mix with just like from the Java and the spring framework world where similar things are going on? Well, interesting point. So, you know, that is quite a myopic view of the world, isn't it? It's the Java view of the world, you know? So, uh, yes, yes. You know what I mean? Which doesn't really reflect reality greatly. So although spring has a whole bunch of tools that are very easy to use and very easy to stitch together, they, you know, you're more likely to use them in an environment that is predominantly using those technologies.

And that doesn't describe everybody. So I think the beauty of the sort of sidecar model, um, particularly was that it abstracted a lot of the details away for things like routing and security and various other things, extracted them away from the software developer who wants to concentrate on business logic, which is mostly what they're paid for in enterprises, right? So, um, I think that the, the, the model is very strong from that perspective. And what's changing here. And like Ed said, it's not just, um, Istio who are talking about this or other brands of, uh, service mesh that are also discussing it is just making the impact of that lighter because it, it, it scales expensively. If every pod has a sidecar with it, then, and you have, you know, 2000 pods, you've got, you know, 2000 sidecars. So if you can reduce that down to instead having a sort of a node sidecar instead, then you only have one sidecar per node, then the volume of those things is, is much less. And the cost, uh, to run that stuff is much cheaper. So from that perspective, from certainly from a cost perspective, it makes a lot of sense. And I think they're quoting sort of 75% in a typical scenario in terms of, uh, cost savings, uh, where they were to, to where it would be in this new model. Um, so if you have a plot polyglot environment and you've got lots and lots of, you know, disparate apps running perhaps on very different technologies, um, then it sounds like a really good way to, a good thing to investigate for your architecture because it will not only save you cost when you build your software, but it'll save you cost when you run it as well. So I think it sounds like a really good innovation. I'm looking forward to hearing more about it. Yeah. Yeah. You know, I mean, you, you do point something good out there that, well, I mean, many good things, not just, but, but that, you know, I, I, I guess, I guess you do always have to

be careful of like, you know, the sidecars wagging the motorbike and making sure that they don't become the act, just like governance, right? Like you don't, you don't want being, being compliant to actually be the deliverable that you're doing with software, which I think a lot of people fall in. And then, yeah, you know, my, my, my facetiousness aside, like, I think it feels like, it feels like, like at a service mesh layer. Well, first of all, I think it bears explaining what the point is, which, and tell me if I'm wrong here, but it's essentially like with the cloud native application, you basically are sticking parts of your application into containers that you have to sort of coordinate and run over a network to make it represent like a whole network. I mean, a whole application. So instead of having just like one process running, you've broken it up into little pieces that like communicate with each other over a network. And then once you have a network, you've got to like manage the components having, being able to find each other and then manage just like connecting them together. And then, you know, to overload it a little bit, you get the chance to do sort of, um, uh, positive man in the middle, not attacks, man in the middle positivity. Augmentation. There you go. We're, we're, we're, we're security wise, you can start to monitor what's happening between this and make sure that it's all good, uh, instead of bad. Well, I mean, I mean, augmenting by adding stuff, you know, adding security, right? So, so your, your end point may in the pod may be unsecured because you can, because the traffic coming into the pod is only dealt with by the sidecar. So the sidecar is a thing that gets secured. So it, it does sort of relieve you of a lot of responsibilities and a lot of, uh, messing about. It just gets in the way of you writing your business logic and, and instead you're dealing with all these, um, cross cutting concerns instead. So, so yeah, it's a useful thing.

I definitely, uh, I definitely think it's useful. So, and then, and then as, as, as, as cool thing, the developers get to throw over the wall to the ops team, we don't have to worry about that, that, that, that is, I mean, that's another trend we should open up with, with hand wavy abstractness in some episode is like, it feels like the direction that the, the Kubernetes standardization is going is to allow us to put that wall back in place, you know, maybe that it was never been removed, but to, to, to make peace and, and, and, and, uh, what was the word you all use instead of man in the middle and, and, uh, augmentation improvement, uh, of, of the wall between the business logic and the operations logic. Uh, I, I, I see it slightly differently. I would say it gives an opportunity for devs and ops to go for a drink instead of going for a meeting about layer, whatever security, right? Yes. No, no dev wants a meeting invite about any of this stuff, right? So, or maybe someday I'm speaking, you know, uh, I'm, I'm perhaps projecting a little bit too much, but I certainly don't want an invite to a meeting, a three hour meeting about that stuff when you could go to the pub and have a drink instead. I, I think, I think if we, if we workshop that phrase just a little bit more, we'll have some snappy fun phrase that you can start using when you're doing your, uh, you know, one minute segments on broadcast TV about, uh, about what we're up to, you know? It's been on bites. Yeah. Deb, drink ops. That's what you're talking about. Yeah. Yeah. We, we, we, we, you know, then, then you can go have a pint together instead of having a PowerPoint together. I don't know. Like I said, we got to work on it. We got to, we got to figure it out. All right. Well, also, also, uh, you know, looking over the news, uh, uh, I think, you know, uh, so, so get labs big, big company that IPO'd recently. They had, uh, uh, their, I think their, their, their Q2, I don't know what their fiscal year

is, but I think it was at the end of July, their, their Q2 report. And, you know, I looked through it briefly and I gotta say, it looks like they're doing a good job. Like they seem, they seem to be doing pretty well there. And I thought, uh, you know, it would be a good chance to kind of check in on them and, uh, and, and see, see what, what they do now. Now, Ed, you follow the space all the time. What, how would you characterize what, what Git labs is into? What's the kind of category that they work in? So I'll be quite honest and say, I've not looked into Git lab too much. The only, so the lasting thing I have, if anyone says Git lab to me is that everyone just wanted to migrate to Git lab when Microsoft bought GitHub, that was the one thing in my head, right? Well, as soon as Microsoft announced it within days, forums were full of, I've had it. I'm off to Git lab. And that, I don't know if that genuinely sort of spurred their business a lot, but it felt like their moment to grab and go, great. We're here. And they've kind of been there ever since. And that might be completely wrong, but that was kind of the only thing that stuck in my head. And beyond that, I haven't really looked into them, I'm right. I do spend my time in this space, but I've been slacking clearly. I've had the fortune of working for businesses that both use GitHub and Git lab. And Git lab was extremely popular, mainly because it had the ability to be installed on print, right? And you can actually put it inside your data center, back to data centers again, and you could have all your own code and devs will get the same Git experience that they're used to elsewhere, but they could have it sort of in a controlled environment, if you like, where they're building their private libraries inside their enterprises. So it was very popular already from that perspective before the sort of announcement of Microsoft buying GitHub, which I'm sure did accelerate it somewhat as an alternative. And, you know, I think the main, the high level deal, so to speak, they seem to have is like, it's an integrated, I don't know, ALM collection of tools, right?

More than just Git as version control on its own. I don't know what the revenue breakout as far as like, which parts of the product line get used more. But they seem to have all of those tools from issue trackers, to wiki things, to your source control, to on and on and on, that you would use in that area. And, you know, I mean, I think there's a lot of us in that area, we want to improve that tool chain there. And so it is, it's always nice to see other people succeeding at it. And I by no means take advantage of the following as much as I would like to at all. But they are, I think, I think the, they're, they're, you know, I guess you would say an open source business. But the interesting thing that they do is, I think they basically do most all of their internal stuff on a public wiki somewhere. Like you can go read the, the quarterly market plan, marketing planning and OKRs and things like that, which is kind of fascinating to, as I say, I don't actually go look at it that much, but it's a fascinating potential. And they're, and they're remote first as well. You know, they've been remote first for a very long time and their HR manual, for example, is one of those online and you can find out all about how they operate as a company, how they operate their IT, how they, you know, support the folks that they have, you know, working from home from places all over the world. One of the most interesting things that's in that manual to me anyway, is that they don't buy windows. So they don't supply staff with windows laptops and stuff. Oh, right. It's like, no, we're not doing any of that. We want a different approach to our network and the way that we, the way that we work and the way that we support folks who are working from home. So very interesting that they've turned their back on Microsoft windows, which is kind of like, you know, the mainstay of enterprises everywhere. Now, I mean, the major consequence of that is they're probably their pivot tables aren't

going to be as great as other companies. Cause I mean, everyone knows that if you want like the best Excel tools, you have to run it on windows. So there, there might be a few tells in their analysis. I can't say anything about office suite. Right. But windows operating system is definitely in there. It's like, no, you can have Apple or you can have Linux, but. So it's actually like policy that about what you can run. Ah, I was going to say, cause I don't even know if you have to pay for windows anymore, but I see if it's kind of, yeah, if it's kind of against networking policy, then that's interesting. Yeah. But there is a, there is a license for the desktop. So yeah, they, uh, they don't want to pay that license for the desktop and they don't trust it. They think it's not very secure. I think from judging by what they say in the manual and, uh, they want folks to use, um, Apple or Linux, which, uh, I'm very proud of using Linux. So I'm, I'm all for that, but coming back to what they actually do, I think where they've really grown in the last few years and, and it's, it's been sort of surprising, I think it's particularly to, um, you know, folks in the industry is, it's just covering that whole secure sort of software supply chain. Like you said, you know, the CICD, the, um, static, uh, security testing, the dynamic security testing, all sorts of stuff that's gone into this huge sort of pipeline of stuff that can help you get from, you know, uh, an idea through to a piece of deployed code. And I think that's very commendable. And that's possibly, um, why we're seeing such good results from them financially as well, because they seem to hit a nerve, you know, that's good as do we, you know, we, we also think that that's a really good thing to have. We're very strong supporters of the whole idea of a flexible software supply chain that's secure. Well, speaking of security, the, the, uh, the sneak state of a cloud security report was out. I think, I think, uh, you, you enjoyed pouring through that, right, Ben? Is that... I did have a look.

Yeah. Yeah. And, and before we get to that, you pronounce it sneak, right? Is that how, how you, like, that seemed, that seems like, I always thought that was an odd choice for a security company name, but you know, it's still pretty good. Yeah. Yeah. I know what you mean, but yeah, sneak in and, uh, uh, you know, an interesting spelling as well. I think, uh, I think some folks get caught out from that from time to time. I've definitely seen PowerPoints with the wrong spelling in there. So yeah. Yeah. Well, what, uh, what, what, what do they say is going on? What's the state of cloud security? Well, yeah. So they've done this survey. I think there was 400, uh, respondents working in the, um, cloud security arena because of course they've got, uh, products in that area as well. Um, I'll give you some of the highlights if you like, uh, 80% of organizations have experienced at least one severe cloud security incident in the last year, such as data breaches, data leaks, intrusions into our environment, 80%. So, you know, four out of five, that's a lot. But also what's interesting is that, um, there were some findings in there about who's most susceptible. So most susceptible, they say in this survey, uh, uh, things like public sector organizations, which you can kind of understand because you know, the, the, the, the, perhaps the investment is difficult to come by if you're, if you're limited in budgets and that sort of thing. Um, and also, uh, what does it say public sector and, um, the best were, were actually enterprises, you know, the sort of folks that, that we deal with. So, you know, who knew if you, uh, if you spend on good tooling, uh, you get better results. So that's a, that's a good finding I think from that report generally. Um, and I think one other thing I'd like to call out, I reckon was that, um, 77% of organizations cited problems with poor training and collaboration as a major challenge.

Now, I like to mention this because I did a video recently about what if your platform actually trained your people. So go and check it out on YouTube if you haven't seen it and we'll put a link in the show notes, but, but yeah, you know, if your platform helped train new people, maybe that can reduce the burden that's causing 70, 77% of folks to have a problem, you know, getting the right skills and being able to solve this security issue. Well, I think this is a good chance for what hopefully will be a reoccurring segment called let's check in with YouTube star Ben and, uh, and see how, how, uh, how, how, how is your YouTube strategy going? How were the shorts? Let's, let's do a review of that. Or do you think it's not as, it's not as regular as perhaps you made out there. Well, I, I feel like it's been a long enough time you've had since last, we talked about this. You've had a couple of months to analyze what's been going on with your YouTube stardom. Well, what I did have any lessons learned now, what I did find out Coté from doing a very, um, small scale AB test was that, um, videos that actually contain my face do worse than videos that don't, uh, completely booking the trend of, okay, we'll have a human in the picture. Some humans, perhaps not. So, uh, so yeah, I take that away as a key thing, uh, for experimentation. There you go. That's a free one for the, uh, the people out there who are working in nerd space with computers. Try, try, try not having so much of a face focus on the terminal. That's right. Maybe, maybe if you had, if you had your head and on your forehead, you put like the terminal, like do a little green screaming, green, or screaming too. But you know, that's where you had the actual code. Uh, well, there's plenty of room. Uh, I reckon I could probably do that, uh, this stream, basically the least good thing possible, isn't it? It's just free talking places.

Exactly. Improved. Well, it's, it's increased by 33%. So yeah, there's a, there's a, there's an interesting thing. Have you ever done an, an AB test like that Coté, cause you, your, your shots are excellent. You know, I, everybody loves them, me included. I, I, I have to say that over the years, I've done lots of AB testing. I just have never analyzed the results of the testing. I, I always, I'm always intending, like I'm going to try something new versus something else. And so I set it up and every now and then I even set a reminder to go check in on something, but then I, I don't. So, uh, I, I don't do that analysis. I'm very, very waterfall in, in my approach. I just deliver the thing and then that's it. And what, what did you say in the preamble? Something about you become very good at avoiding. Ah, yes, yes. There's, there's, there's, there's a, there's a lot of things I should be doing, but I've become really good at not doing them. And, but you know, or, or, or maybe I, or you could also say there's a lot of things I should be doing, but I've been very busy not doing them. Uh, you know, is another. He prioritized what not to do. That's very, very admirable. Exactly. Well, uh, finally, you know, speaking of. Of reports and surveys, there's the, uh, also the new Relic observability report came out and, uh, I knew you took, I know you took a look at that, Ed, what, uh, what jumped out at you from that? So the first one, and I think to be fair, this is kind of every survey, right? This is probably everyone getting them out and ready for cube con and things. Um, the new Relic one, a bit like the, the SNCC one said, Hey, we do security and everyone needs security. The new Relic one said, actually, everyone needs observability tools. Um, there's some stats. The first one was quite a big survey, actually. They had over 1,500 sort of people fill it in. Um, but they ended up saying, I think, where was it? Um, 82% use four or more observability tools, but basically everyone is using something. Um, and that to me, I mean, that's absolutely no surprise. I would have probably put that at a hundred percent. Frankly, everyone, as long as I've ever been in it has wanted one tool to monitor everything.

Um, and no one has ever had that and no one will ever have that, um, they did say in their reports that 2% did say they did it all with a single tool set. Oh, the 2% of probably lying or just unaware of all the other tools that they actually have. Yeah. Yeah. But it's no surprise to anyone, right? There's a lot of companies in that space. So lots of tool sets, some of which we do as well. Right. Um, and probably not the only one you have, um, or it's a, or it's a small business, right? That, cause that came out in the sneak report as well. It's like, Oh, surprise, surprise. It's easier to be secure if your footprint is less. Ah, yes. And I'm sure the same is true with the deserveability, right? If you've only got a certain number of technologies or vendors, whatever you probably can do with one of these unified tools. Yeah. Most companies aren't in that position. And then you're just like, you know, the minute you get above a certain level of detail, the all in ones never do the job quite right. So not a big surprise in that respect. Um, they did have a, quite like a third of organizations rely on manual checks and customer complaints to identify interruptions. And again, I'd have put that number much higher because however much monitoring is somebody will always either complain or find something you haven't noticed either way. It was an interesting approach of actually, for some people, right? That might still be the most cost effective way of doing it. Customers complain. That's fine. That gives you a quick alarm. You fix it. Maybe you don't lose revenue. You lose a bit of customer interest, but, um, for some, that'll work. Um, blockers for adoption of them. The biggest one was lack of understanding of the tool sets. I think, and what you need to do to get full stack observability, um, which again, wasn't too surprising. There's a lot of complexity in there. And I think people used to just think, especially from the IT sort of infrastructure side, well, monitoring is monitoring. I, I, I know what that is and the new world of observability is a long way from that. And there's a lot more business understanding required as well as a lot more sort of technical tooling. Um, so that and budget, having the budget to actually get the observability tools. I mean, now you have to understand what it is you need, and then you have to be able to explain to people and give them business outcomes and implement it in a way that works.

And doing that to budget obviously is a bit of a challenge. And they did point out also that the, the companies who felt they had full stack observability had put something like 5% of their revenue into getting that tooling. Oh, wow. It's a significant chunk. And I think a lot of companies would struggle to justify that, or it would take them a long time, right? They'd go through that iteration. Um, you can't just turn up out of blue and say, we've got all this lovely new cloud native stuff. We're migrating 10% of our IT to the cloud, whatever you're going to call it. Oh, and by the way, we have to spend this much observing it all. That's a hard ask for people until they realize that it actually is worthwhile. So yeah, those are, and the only surprising thing I saw in their survey that really stood out to me was of all, now this was from New Relic and sort of, I guess they've categorized it the way they do things. They cover observability for everything, but it was interesting in on the world of these Kubernetes came pretty much last in the list of things that people wanted observability for. So they had 20 other categories and it was rock bottom. Um, and you know, on top of that was sort of, it's a little difficult without seeing all the questions, network monitoring and sort of database monitoring and infrastructure monitoring. But what they saw as different categories, right? So presumably you would not be monitoring on Kubernetes network. It would be generic network monitoring that came top. Kubernetes as a specific area to monitor was very low down and they, they put it down to just, well, actually on the scale of everything, coming back to how much IT is in the cloud, Kubernetes is still a small chunk. Um, when you look at the whole landscape, whether that's mainframes, x86, everything, Kubernetes is new, is well, you know, new depending on your audience, um, still being adopted on that scale. Yeah, I'll have to check that out. Cause I mean, all the way back to our discussion, right? Like looking at, I guess it's a, uh, here, correct me where I get this wrong, but it is a trailing indicator and, you know, something that's happening in the, in the past of basically to look at what management companies are doing like new relic and others. Like it's a good indication of the, this types of it being used.

It sounds ridiculous to say it out loud, but you know, it's, it's, you generally manage some type of it that you use, hopefully. And so to see what that breakout is, it gives you a good indication of like on-premise versus public cloud or what database or this, that, or the other. And new relics have been around long enough that they probably have, I mean, they're not like a custom built Kubernetes observability startup. They, they, they aspire to manage everything. So they have the, uh, the full gamut there. Well, uh, if you want to find links to everything we went over, you can go to tanzutalk.com and sadly you will have to find this episode. But if you're really interested, I, I, I wager you won't be listening to this six months or two years into the future. So if you go there within this week, you'll probably find it on the top and you can click on it, uh, you know, which seems easy to do, but also, uh, you want to tell people about a conference we have coming up, Ben? How, how would you describe this thing? We have December 6th to 8th in San Francisco. What's going on there? Uh, six to eight in San Francisco. Is this December? Sorry. Did you say December? Yes. I thought I was going to challenge you to pick the month. This is the SpringOne, right? That's right. You got it. SpringOne. Yes. Coming back in person. Holy in person event. I understand this year. And, uh, we'll be at the Moscone center in San Francisco. Uh, whole bunch of interesting, uh, talks going to be there and speakers going to be there. And, uh, yeah, it's going to be a lot of fun. And I believe it's possible to get a discount on your ticket, right? Coté? That's right. If you use the code Coté 200, it will get you a mysterious amount of money off of your registration. You'll have to go fill out the registration and put in the code Coté, C O T E 200. And you can then see how much of a discount that you get. It's a, it's a fun little game you can play. Uh, hopefully it shouldn't be missed. The, the, the keynotes are always brilliant.

You hear from some of the best folks in the, in the software industry who are using, uh, spring and anger. I remember a few years ago, for example, it was, um, uh, Blizzard, you know, the games folks were on the main stage, which was, which was really cool. Uh, so yeah, don't miss it. And also the, the after parties and the events that go around, around the same time are also brilliant, you know, really, really good. So yeah, sign up, do it. And I, we're, we're, we're planning a, uh, podcasting and streaming booth that we're going to have booth is at, let's say area. And so I think we'll have some Tansu talk stuff done live from the floor. I mean, hopefully it'll be on the floor, not levitating somewhere. Uh, so we'll, we'll, uh, we'll figure that out. Well, uh, well, I've just got to say, I thought you were going to ask me about VMware Explore Europe. Oh yeah. That's the first time that Explore is coming to Europe. We've never had Explore in Europe before. So don't miss that that's, uh, November 6th through 10th, right? No, 7th through 10th, 10th, maybe. Um, yes, uh, Barcelona and, uh, I'll be there. Coté, you're going to be there. So yeah, don't miss that either. That's, uh, you know, a different sort of thing, I guess, you know, it's a broader sort of set of, uh, of use cases being discussed, but we'll have lots of customers, lots of partners, loads of interesting stories and topics. I think, uh, if you come down to the stand, you'll be able to certainly meet me and, uh, possibly you'll even see a beardy Coté wandering around. Definitely. And, you know, it'll be great because, uh, you know, if you're interested in the Tanzu application platform, it's a great place to get all the tapas that you want. You can, you can definitely, uh, uh, that was, that's my joke for the episode and we should end before I make any other terrible ones. Well, if you want to get the show notes for this episode, like I said, you can go to tanzutalk.com and, uh, remember you should definitely subscribe to this podcast. And even if you don't want to listen to it, you can subscribe to it. Find all the podcast downloading devices in your house from your, your loved ones, your enemies, your children, open up the podcast app and subscribe to it.

You can set it up to auto delete. It needs to download the whole episode every week. Doesn't matter. But that's, uh, that's what I recommend for maximum enjoyment in life. And with that, we'll see everyone, uh, you know, next time. Bye bye. Bye. Bye.

Tanzu Talk · 🍕 Get more pizza.