ia-Redmonk-CloudComputingAndLinux675-3

40:40 · SRT · archive.org · all RedMonk TV

Hello everybody, this is Michael Coté with RedMonk. I'll be giving this talk tomorrow, what is it, April 9th? I probably have my date wrong because my watch is always chronically wrong, at the Linux Foundation Collaboration Summit. And as I was rehearsing this talk, I thought, why don't I record it so I can publish it somewhere. The idea of this talk is to give a brief overview of cloud computing and talk about how, obviously being at the summit here, how it relates to Linux. And to that end, here's some expectations or an agenda, if you will, of what we'll be talking about. And for the people listening to the recording, this is supposed to be 40 minutes, so if you find me rushing along or something, I'm trying to shorten what's probably an hour or more long talk into just 40 minutes, where there's a discussion afterwards that people will be talking with. So, getting back to the track here. The expectations here is that, you know, when we come to cloud computing, to me, cloud computing is obviously one of the larger hype tech things or tech hype things that's out there nowadays. Especially in the sort of gas bag world that I'm in. Or, you know, however you want to describe whatever crazy thing it is that I do. Basically, you know, it's what people are always interested in talking about. And in that sense, it's become what SOA was like very early on. It's sort of a silly putty, if you will, that you can jam into any sort of problem or any hole that you might have to sell to someone. And I tend to take a pretty pragmatic take on lots of these technology things and go with the most simple definition possible just to sort of clear the deck and get on to the more interesting things, which usually tend to be the consequences and how it relates to existing stuff. And to that end, I'm going to go with a pretty simple definition of cloud computing. And since, you know, the focus here is how Linux fits into the cloud computing world, I'm not going to spend a whole lot of time debating with myself, so to speak, what exactly cloud computing is or its importance or things like that. We're just going to set it aside as an interesting sort of touch point and see how the, you know, Linux fits into that. And to that end, you know, one of the easiest ways to go comparing and seeing how a new technology or an idea sort of works or doesn't work or fits in with an existing technology is to do what they call a SWOT analysis,

which is basically just the strengths, the weaknesses, the opportunities, and the threats. And that'll be the, hopefully, the bulk of the talk that we go over here. So first off, we need to establish a baseline of what cloud computing is. And as you can see on the, with this diagram here, this pretty excellent diagram from Christopher Hoff that was published a couple months ago, you can really just go crazy with what you think cloud computing is and get like a really complex, detailed definition. But when I think about cloud computing nowadays, for as long as I've been sort of, you know, excited about it, been a cloud nut, as I like to say, I end up nowadays dividing it into what I always call, you know, being, you know, not being able to resist a good joke, the three asses of cloud computing, and that software as a service, a platform as a service, and an infrastructure as a service. And one of the, my other, my fellow Red Monk, Stephen O'Grady, he makes a differentiation between SaaS, between SaaS essentially, instead of calling a platform an infrastructure, he calls it fabric versus instance space, which is another way of breaking it up. So as the diagram shows, obviously, there's a whole lot more that you could say that cloud computing is, and a whole lot other ways you could divide it out. I mean, people will just go nuts and have storage as a service, and security as a service, and networking as a service. And basically, every single box and more that you see here in Christopher's diagram as a service. So the question is, you know, looking at these three asses, what are, what are these things? And again, to use a very simplified definition to sort of move the discussion along, what I look at it is that it's another way of describing the classic sort of triumvirate of computing of IT that you have. And there's basically applications, middleware, and servers. And, you know, the ordering there is more of a layering on top of order, not like a judgment call about the importance. But you can see, obviously, that software as a service is actual applications that an end user interacts with, whereas platform as a service or the middleware stuff is just the technical goop, if you will, that you use to build together applications.

And then all of this, of course, has to run on something. And, you know, in the physical non-cloud world, if you will, non-virtualized world, those are just physical servers that you have, where the servers are obviously composed of operating systems and the hardware that's backing them. And again, I'm greatly simplifying everything here. I mean, there's obviously networking and storage and security and all sorts of other things that come into play. But I think, you know, when you're looking at something as very hype-fed and new of an idea as cloud computing, it's important to keep things as simple as possible, especially when you're interested in other things, other than just sort of, you know, how many angels you can get on the head of a cloud, if you will. So moving on, you know, you know, it's always important, I think, when you have a technology to figure out why you should care about it. Because there is, I mean, the technology world is really just a bunch of problems looking for solutions. Or I see I even got that wrong. It's a bunch of solutions looking around for problems. I've been so well trained to think positively about that little trope there. But I think, you know, unlike with sort of, for example, with SOA and a lot of other big hype-fed technologies and sort of practices in different worlds, I tend to have a lot more hope for cloud computing. Because the benefits are a lot more, you know, there's the phrase business-oriented, but they're a lot more goal-oriented rather than necessarily technology-oriented. There's not a lot of people who sort of are necessarily at the wonderment of the dancing bear that cloud computing is. Instead, people tend to be excited about it because of what they can do with it, what goals they can achieve. And to that end, if you divide the world up, as I tend to do, into users or buyers and vendors or software sellers or people who are producing software, they don't have to necessarily be selling it. It could just be an open source sort of thing. But these are, like, the primary concerns, you know, the reasons why people tend to like cloud computing. And for users, at the moment at least, and depending on the sort of burstiness versus 24-7 runtime you have, cloud computing is extremely cheap compared to maintaining your own on-premise things.

And part of that cheapness is more than just sort of the charge that you're getting for it. But it's also driven by the flexibility that you have with cloud computing. I mean, all the startups that I talk with nowadays, to overstate it, obviously there's plenty of startups who aren't like this, but many of the startups are people starting things up. What they really like about cloud computing is how fast and easy it is to get up and running and change things around. It's a lot better than, you know, going down to the data center or messing around with physical stuff in your own data center. And it's that quickness, that flexibility, that malleability that users really like quite a bit. And, you know, all the high-flying cases of cloud computing, they always talk a lot about the elasticity. I mean, that's the E in EC2, if you will, and the scalability and being able to handle, as we used to call it, being slash-dotted or being digged or whatever. Just handling more loads as needed on demand instead of reaching that capacity limit that you have. And as I think we'll get into in the rest of the talk here, you know, this elasticity is not really a majority concern, I would say. It tends to be something that only a select amount of people want to use. But more importantly, if you look at elasticity and scalability as the introduction of sort of unlimited computing power at your disposal, that's where things can get kind of interesting, where you have access potentially to a lot more parallel processing and larger batch job processing that you might have. So for users, you can see that there actually are, you know, the, again, it's the effects of cloud computing that users tend to get excited about, not just the ticks and the flows and cloud computing and the wizardry of it on its own. And for vendors who are very excited about cloud computing, I mean, the primary thing that cloud computing gives them is brand new business models to sell software to. And any technology vendor is always going to be excited when there's a new product that they have to offer, when there's something brand new that they can start selling. Because eventually with any technology, it sort of taps out and you start getting maintenance fees and you need to come up with something new to sell. So new business models are always exciting. And, of course, what drives this is there's new features that people can add, that vendors can provide.

So whether they're actually selling something new or they're, you know, adding new features to existing stuff, it's just another product that they can sell. And then in theory, you know, cloud computing hasn't really been around long enough, but in theory, you'll have lower cost of ongoing maintenance for your application. So, you know, looking at the very top of the stack when we get software as a service, instead of having to update and install all these different on-premises versions of your software, you'll just have the one version that you can get up and running. And, you know, of course, the granddaddy example here would be Salesforce. And one would hope, again, in theory, that there are some economies of scale by pooling things together. But we'll see if that actually pans out. So I got to some of these, but here's just a quick run through. You know, it's always good to, like, look at the concerns around a new technology, the potential problems that we'll have. And top of the list is, especially when vendors get involved who want to do enterprise sales, is there's a lot of semantic confusion. I think when we're talking about cloud computing and people don't really divide up their thinking of public versus a private cloud. And a public cloud is what people usually focus on when they talk about cloud computing. And that's essentially running something on Amazon or running it on forest.com or running it on some cloud.com on the public internet, accessible in someone else's data center. But also when people are discussing the cloud, they also might be discussing private clouds, which means an organization or a company is going to have their own cloud behind their data center, behind their firewall. And they're going to do something with that cloud. And for, you know, if you think about it, there's, there actually are different scenarios and concerns and things. If you're actually, if you own your own cloud, it seems like there's a bit of missing the whole point in the first place. So it's always important to separate public versus private cloud discussions when people are talking about cloud things. And another concern was, I was saying that, you know, companies will have at their disposal a lot more processing power. And I think most developers, your run-of-the-mill developer, myself included when I was a developer, are not very good at multi-threaded programming or, you know, concurrent programming and all this, you know, batch-oriented things.

And the style of programming that cloud computing and things like Hadoop and whatnot sort of encourage you to use in the infrastructure. I mean, to scale out, you have to think in parallel, which is not a normal way for developers to think. And I think on the other side of the house, when it comes to running clouds, the operations people are not, they don't usually have very grid-oriented sort of thinking, very cloud-oriented thinking. So there's, the whole point is with a new way of, a new paradigm, if you will, to use a quarter word there, there's all sorts of retraining and reprocessing that you have to do with your existing IT organization if you're moving on from a more traditional setup. And that's always a concern. And the other concern is, I was alluding to, you know, that the cost of cloud computing is low, but usually, for example, on Amazon EC2, if you do the comparison of running something 24 by 7 versus running something just in a bursty way, you end up paying quite a lot of money. You could just pay a hoster to be up all the time or have your own infrastructure. So it is always important to be leery of ending up paying more when you're just using cloud computing. And this is a bit of an exaggerated example, but I was talking with someone who manages, excuse me, manages a very non-technology company that has a bunch of franchises and retail locations, their little branch locations. And they were talking with one of their HR providers, their ERP providers. And the ERP provider comes in every quarter and wants them to move to an off-premise cloud-hosted SaaS thing. And they're going to charge them, you know, the low price of $150,000 a year to run their payroll stuff, you know, in the cloud, if you will. And so this guy did the math, and it turns out that he was either running it in-house and they were only paying $20,000 a year, you know, they were paying for a Windows license, a Windows server's license, and for the actual hardware itself. And then, you know, for 10, 20 hours a year of a DBA to come in and, like, muck around with their stuff. And so, you know, for him, while you might look at cloud computing, it's like a great cost-saving thing. Like, actually having something on-premise was much cheaper. And then there's also, at this point, with any new technology, I mean, there's lock-in concerns that you'll have that things aren't portable between different clouds.

Same concerns you would have with operating systems that you might be locking into or whatever that middle layer is, if you will. And then, of course, there's another concern is looking at transitioning existing legacy concrete, if you will. And while cloud computing might be a more efficient, better way of doing something, it just may be the fact that transitioning what you have that's already working, that may cost a lot of money might end up not really giving you a good return on investment. And you might end up with just the same sort of expenses and the same problems if you move your legacy stuff around. So you don't want to be – you want to be – as someone running IT, you always want to be conservative about mucking with something that's working. And I think, you know, looking at the wide context of something, I think an interesting thing to worry about is that getting obsessed with cloud computing might sort of blur the importance of virtualization in the overall world. And virtualization, of course, is a huge part of cloud computing. But it could turn out that cloud computing really isn't that big of a deal. And what we should have been more interested in worrying about was virtualization this whole time. And I don't have it listed here. But another thing that always irks me personally is, you know, I think right before cloud computing was a big hype concern, we were all talking about software as a service and SaaS and all these great things. And I actually think that for the mainstream world, SaaS is a much more interesting prospect than cloud computing, if you will. And again, SaaS can be part of cloud computing, but I think focusing on the applications that you're running is really the more widespread, long-term thing that people should be worried about, not necessarily the full stack of cloud computing. So for something as new as cloud computing, it's actually – as sort of an industry watcher, it's actually very handy because you're still at the point where you can kind of point at things and say, that's cloud computing. And thankfully, at the moment, it's ending pretty rapidly here. But there's a small handful of people, a small percentage of those that handful, you know, with logos up here. But basically, you can – there's good examples you can point to of things that are, you know,

the infrastructure and the platform section. And I didn't list SaaS here because I think SaaS things are pretty obvious, like Salesforce itself and Gmail and Google Apps and Zoho and all these things are all just software as a service or web applications, as we used to call them. But you can look at Amazon EC2, which is an instance-based thing, to use Stephen O'Ready's term, but it's infrastructure. It's just raw compute that you have. There's not really any middleware on top of it. And, you know, S3 is part of that. It's just raw stuff that you have access to that you need to layer on top, manipulation and usage on top of. And then looking at the platform stuff, Force.com, which is Salesforce's spinoff of its own middleware, if you will, is a platform where there's actually services and APIs and things that – sort of things that you would see from the Java enterprise world or the .NET world where there's established buses and ways of doing things and APIs. And I'm being very imprecise in my phrasing because I was always trained to just think of services as a developer. And Microsoft Azure, though it's not out yet, is trying to – is being the same thing where, you know, you might have identity service. There's just basically middleware that you can plug in to use. And there's other things to look at, like Sun has the – the cloud that they're working on and they have some nice demos of. And 3Tera is a provider of public cloud – I mean, of private cloud infrastructure and they have some public cloud partner hosting. And Rackspace has been doing a lot of cloud things by acquiring Slicehost and JungleDisk and a bunch of other places. And they have their own Maso. And there's a bunch of hosters that are sort of rebirthing themselves as cloud providers. For example, there's Opsource that has been around for a long time. It's more of an ASP sort of layer, but they fit well into the conception of what a cloud is. And additionally, I think there's, you know, in addition to virtualization, one of the other things that drives cloud computing is the – you know, when you have a bunch of different nodes you're managing, there's a lot of automation and provisioning that you have to worry about. And there's folks like Puppet that's sort of patronized, if you will, by Reductive Labs, the company around it, who are, like, very much so involved in a lot of the gorpy-morlock-level stuff of cloud computing.

So these are all just some people – when you're thinking about who's doing cloud computing and you want some examples of stuff that is or could be cloud computing, you can look at these things. So let's jump right into the SWOT analysis of things. And I think we're at 16 minutes and 32 seconds, so we're actually doing quite well in the time. Maybe I can even slow down a little bit. But so we'll start out with the strengths that Linux has. And I think when it comes to cloud computing, all of the strengths that you would think about for Linux and all the distribute distros that exist out there, they apply in the cloud computing case. I mean, top among them, of course, is the sort of potential or perceived or whatever you want to say cost of Linux that potentially could be free for you, monetarily at least, to use. And obviously, as Jim Zimlin was saying yesterday in the keynote, you know, I mean, would Google have been possible without having Linux? I mean, would Google have been able to afford all of the – or even negotiate all of the, I don't know, Solaris licenses or Windows licenses or whatever non-free license they were negotiating for all these, you know, however many thousands of servers that they have or instances? And when it comes to the wide-scale, virtualized, like, pools of resources out there, if you're going to have low cost, something has to be free, essentially. And I think having the operating system, which does much of the coordination here, be free certainly is a huge benefit for Linux, as has been proved out by most, if not all of the cloud providers, or I would say the majority of them, as I would get told on Twitter that I would be wrong about from Sun folks and other people. The majority of cloud people out there that exist now are based on Linux things or using free software. So that's a huge strength that I'm overemphasizing, but nonetheless, it's worth talking about. And I think, like I said, all of these things are the motherhood in Apple pie of why we love Linux. But most importantly, you know, Linux has been around long enough and been road-tested that people know that it's reliable and it works at scale, and it's an enterprise-grade operating system, if you will, which I don't think anyone really doubts that very much.

But another interesting thing, you know, cloud computing is, you know, it's based on old concepts, and we've had good computing and things like that. But basically, at the end of the day, if you're going to be a cloud computing provider, you're going to have to muck with the operating system and have it be kind of malleable and make it fit to what you want, whether that means carving it down or changing things around or whatever. You're just going to have to mess with it. And the way that a cloud provider would set up a cloud, they need to differentiate somehow. So they, in theory, don't want to have the same exact infrastructure, or they would be avoiding lock-in, for one. But then there would be no reason to go with them instead of someone else. And from this respect, you know, in a sort of open-source counterintuitive way, the malleability of Linux is very nice because you can shape it to be sort of whatever you want. And that's kind of due to the transparency that you can actually stare into its guts and understand what's happening. And it's a classic open-source sort of thing. I mean, this is the same reason for a technologist why any piece of open-source software is good, because you can understand it and mess around with it. And, you know, the next thing when it comes to cloud computing, again, it's one of these things where you end up, you know, I always love the classic old saying, you know, we had this problem, and then we bought some technology, and now we have two problems. And when it comes to cloud computing, it's essentially that sort of thing where once you introduce cloud computing or virtualization, as we'll get to, then you have a huge management problem because you've got all this dynamic stuff running around, and you've got to have some way to corral it and wrangle it and manage it. But thankfully for Linux, you know, having worked on this stuff firsthand back when I was at BMC working on the, you know, a bunch of IT management software, you know, whenever we would have to manage Linux, and to a lesser extent, Unix-based systems. But whenever we would have to manage Linux, whatever distribution, it was typically a lot easier to write management software for Linux stuff than it was for proprietary and closed things, especially for Windows, which at the time, at least, was very much, you know, I don't want to get into it. It was very much more so a binary thing rather than flat files and other things that were very easy in Linux. And part of that comes from just the general, like, you know, the pipes and redirects and the plain text and everything, sort of the philosophy of Unix programming, if you will.

But that carries over into Linux quite well, and it's a lot easier platform to do tooling for if you're doing weird, kooky things with it. And so the next thing, like I was saying, is virtualization is extremely strong in the Linux space. And there's numerous, you know, open source virtualization things, and, you know, like KVM, as the previous speaker was talking about. And it just, you know, it's one of these things that works well and is understood, and it is itself open source and sort of, you know, very recursively benefits from all these same sorts of things. And virtualization, of course, is extremely necessary for cloud computing. Otherwise, you just have magic provisioning and, like, what was that old thing, ghost or whatever, that would, anyways, getting off track here. And so, you know, I think the other thing that's interesting, and I'll sort of contradict myself when I get to a weakness in the next slide, is that there are a lot of applications and existing software that's already Linux compatible. And now, you know, one should be never be naive enough to think everything that I've just said about the malleability and being able to, like, you know, silly putty up Linux to stick it into the cloud hole that you have. Obviously, if you're a good QA department, you're going to want to retest all these things that are supposedly compatible with whatever distribution you're concerned with. But, in theory, if you're basing, you know, your cloud computing infrastructure or using a cloud compute, you know, a cloud that's based on Linux already, it should be easier. And I say should with, you know, if I had multiple tongues, I'd be in both of my cheeks there. But, essentially, it should be easier to run these applications on top of Linux-based clouds, which is a nice strength for Linux to have. So, on to the weaknesses, and here we go, like I said, I was going to contradict myself, is that it's a little hard to sort of explain this. But cloud computing, to me, again, is all about the goals and the effects and the application layer of things. Whereas Linux is more of a Morlock concern, like, way down at the bottom of the stack. And they may be eating all the pretty blonde people on top and everything, but that's where the attention is, is at the, where they be a Lodi or whatever.

And, you know, I think the weakness here is not so much destructive, but it's that, you know, it comes in the form of a threat that, you know, Linux could just be sort of like another piece of the cloud stack instead of being a dominant part of the stack. And the attention could move, and for me, hopefully will move, to the application layer, the SaaS layer, where people are concerned about applications. And there isn't really that much of affection for what's going on underneath the hood, if you will. And that's just a matter of sort of like the overall care that the mainstream has with the Linux community. And, you know, I've been around long enough to know that when you make a statement like this, there's all sorts of very fast corrections that come in and correct you. So that'll be a great discussion point to figure out to how I can remove this bullet point from the weakness. And so the next weakness, I think, is another sort of, I'm always overly cautious when I'm telling people what their faults are, just because why not. But anyways, another weakness is that, you know, in my view, Linux doesn't have a huge developer allegiance. I mean, it's just like something that developers, application developers will run things on top of. Now, there's an exception like the LAMP stack, which Linux, being an integral part of that, has a huge appeal to it. And there's certain tools that, you know, the ease of setting up tools and of using AppGit and other things and package management, it's great for developers. But, you know, if you look at the wide array of Java developers and application developers and sort of enterprise developers out there, you know, there's not a lot of people who run around saying that they're Linux developers or programmers. It's more that they're Java developers or whatever framework that they're using. So the allegiance is usually to the framework and not the operating system. And again, in cloud computing, this becomes a concern or could become a concern because your allegiance could be to the cloud that you're running on, not necessarily sort of like the tooling that's within the cloud. And another concern is basically, you know, there is a ton of virtualization out there and there's a lot of automation tools out there.

And as there's more and more fragmentation of different hypervisors, or if there is more fragmentation of hypervisors, and there's already, you know, more than one hypervisor running around in the Linux world. I mean, it's just sort of more hassle and something to worry about when you could just have, you know, one hypervisor and one automation system that you're worrying about. In the same way that, while there's tons of file systems out there, I mean, there's not, in use, it seems like there's, you know, file systems are not really fragmented in the eyes of the users. People don't have to worry about the file system they're using too much. Whereas when it comes to virtualization, it is of a concern, like which hypervisor are using. It's of concern what automation software they're using. And then as always, with any open source project, and because of, you know, Linux is susceptible to what I would call bad citizens, people who may basically take, take, take, and take, and not really do anything to support the community. And there's also plenty of good citizens. But you can see that if we move to a cloud computing world, that there's even more of a chance for people to be more disconnected, which is really what all these weaknesses, the first two weaknesses at least are getting to, is having the community, the end users, not necessarily end users, but people who should have that connection with Linux, sort of losing that connection. And that's really what, I mean, as you look up at the application layer with stack, as I think cloud computing encourages you to do, you lose that connection to the bottom layer of the stack. I mean, you don't have intimacy all the way down to the chip really anymore, because we've lost intimacy. And you could see that the world of people using community would sort of lose this intimacy with the operating system. And, you know, in theory, things could even get to the point where the operating system doesn't matter. And it's just the fabric that matters or something like that, which I think is, is, is a interesting discussion in a song. So all those, all those negative ninny things aside there, or whatever that, that nincompoot phrase is, it's, you know, the opportunities, I think, for Linux are far outweigh any of the threats or the weaknesses or anything. And, you know, I do a lot of, in addition, I have a kind of a schizophrenic, weird coverage areas, if you will.

I mean, we're generalists at RedMonk. But in addition to covering boring stuff like IT management and software development, I pay attention to things like RIAs and Web 2.0 and other things like that. And, you know, the rich internet applications are another solution looking for a problem, like any technology. And I've been, you know, I've been trying to find a back end for RIAs because they sure are damn fine looking front ends, if you will. And increasingly, it seems like there's sort of an aligning of stars happening where all this cloud computing stuff is positioned well to be basically the back end for RIAs. And so it's basically positioned, essentially, we're creating a whole new back end where there could be a whole lot of new front ends on top of it. And one of those front ends, you know, I've only recently gotten sort of like jadedly or jaundacely sort of enamored of netbooks is, you know, you can look at netbooks. I mean, at one point, someone wanted to call them cloudbooks or something like that. And you could see that there, the idea is that you really do have, you sort of like got the physical version of this fantasy of the browser being the operating system, right? Where instead of it being a piece of software that replaces the operating system, it's basically embodied in this hardware and this netbook that you have. It's lightweight and, you know, depends on being connected and depends on the cloud existing. And I think, as is painfully obvious and is almost, you know, I mean, it's just the case nowadays. Tons of netbooks deploy Linux desktops. And I think it's an interesting, you know, cracking the mainstream Linux desktop market, if you will, has always been the holy grail of Linux stuff. And, you know, I thought during the keynote yesterday, there was a nice sort of like, what does a desktop really mean thing going on. And I think this netbook thing really gets to that idea that, you know, there's not really that much reason to switch your desktop around if you don't have a compelling back end for it. And I think cloud computing is providing an interesting reason to care about desktop links, to care about netbooks, to care about RIAs and mobile things, if you will. And I put that in quotes because it just means usually small computer that fits in your pocket.

And another good opportunity for Linux is when it comes to automation over virtualization and cloud computing, there's this sort of increasingly theory of operations that sort of model driven automation, if you will. And it sort of harkens back to the autonomic sort of, you know, fantasy world that marketing stuff that we had in the past that was a great idea. And it just turned out, for whatever reason, not to sort of pan out for everyone. But nowadays, there's this return to the idea that you could actually declaratively model the infrastructure that you want. And there's software that, using virtualization and cloud computing principles, will sort of make it happen for you. And you can look at a lot of the discussion that the puppet guys, the reductive lab guys have around this. I mean, this is one of their big theories. And they've been doing some interesting work with a company called Control Tier that's sort of uniting this together. And then on the very high end, the work that Cisco and BMC, or the BMC is doing with Cisco around the unified compute thing, it's interesting to talk with them about that because they use this model driven sort of approach to doing automation that is an interesting way of configuring things. So how does Linux fit into that? I mean, Linux basically fits in because, again, it's the easiest thing to fit a model to. I mean, it's available for anyone to go out and manage, if you will. It's not quite as difficult as managing Windows might be. And so it's just the potential is there for people to go mess around with it and do a lot of model driven stuff. And the people who are interested in this model driven stuff, they tend to be Linux heads. So it's just the space they're doing things in. So the most obvious opportunity is already being played out. And that's basically that you have, you know, to be all funny about this fragmented airspace. So everyone wants a cloud or, you know, in comparison, you know, you could probably count the people who want to run their own cloud on your fingers. There's, you know, there's probably 10 or fewer of them, but they're all going to want to use Linux. Well, again, I'm overstating it. Many of them are going to want to use Linux or some free operating system. So obviously there's just the continuing just proliferation of Linux out there that people will want to use it. And, you know, it's obvious, but you've got to state it.

And then another interesting, someone pointed out to me when I was talking to them about this topic was that, you know, there's been great strides to make using Linux as easy as possible, to introduce yourself, to bootstrap into it. And you can download, you know, you can download live images and USB images and things that are in comparison to the old bad days of Slackware or just like drop dead simple. Like even I can figure it out. But even more, even more simple is basically just logging into a cloud somewhere, logging into EC2 or whatever and just spinning up an instance and mucking around with it and then destroying it. And this is the most important thing, I think, for the easiness of playing around with Linux. It's like it's one thing to get it up and running. And then when you mess something up, it's another thing to sort of like quickly get it up and running again. And you can imagine that if you can go through and have a fresh instance of something to play around with, you know, 10, 20 times a day, it's a lot easier for you to like make a bunch of mistakes and learn from your mistakes. And virtualization, of course. I mean, this is really just virtualization, but it's even easier if you don't even need to mess with your own personal desktop or your own hypervisor. You just log in somewhere. And so it's just it's sort of as easy a possible way to like play around with Linux. And what's good there, again, is just getting more people interested in using Linux and lowering those barriers even further so they can just get familiar with it and get friendly with it. So what are some threats to Linux? Now, threats here in the typical SWAT thing, they're different from weaknesses in the sense that they're external threats. They're not something necessarily that that's wrong with Linux. They're just things that are happening in the outside world. So I already got to some of this in the weakness part. I mean, there's there's always parallels between weaknesses and threats because you obviously want to threaten your enemy in as weak points, not as strong points. Or your interlocker if you don't want to be all judgmental in your phrasing. But I think there is a danger of, you know, looking at it. And again, it's this is the extreme case, but it's good. It's a good way to think about what the pragmatic real case would be that if there is more focus on the upper layers of the cloud stack on applications and middleware, the question arises, like, why does an operating system matter, right?

Like, why do you have to worry about an operating system? And I think an interesting analogy here, and I forget if they're still doing this, but I remember there was talk or there were these these cars with hoods that you never really open. You don't ever have to, like, go muck with them. And then, you know, it sort of makes the engine of the car not matter so much. Someone was was quipping to me yesterday. They they were working with some some company who wanted to sell the engine instead of sort of like, you know, what you would do with the engine, sell the experience. And that all sounds like, you know, that all sounds like a bunch of like marketing hocus pocus and things like that. But the important thing is that, you know, when when when you go to people who are decision makers or have money or whatever, the people who are going to decide your fate if you're not deciding your own fate, the way they're going to make decisions are informed about what it is they touch and feel and hear about and what they have an intimacy with. And if people are no longer or, you know, very if they're not interacting with operating systems with Linux very much in the cloud computing world, then it's hard for them to figure out the value of Linux. And it just makes the job more difficult, if you will, to sort of sell Linux and drive people's interest in it. So another, of course, threat is that if you have cloud computing, there'll be plenty of people who who won't want to use Linux. So they'll basically not have Linux at all in their cloud. And again, if you have this focus on the application layer of cloud computing, then that means that you're you're sort of the sort of closed non Linux world, the proprietary world has a pretty good backdoor to capture market share again, to capture attention and usage. And you can see that Apple has, you know, I guess, based on on long ago, based on BSD, if I remember, they they've done a good job of basically carving out their own closed source world nowadays, where, you know, you can there's not really an Apple cloud per se, but there is an Apple experience or community that is a very closed world that doesn't really involve as much open source as we might hope. I mean, now they have cups and all sorts of stuff like that, but it's not really the ideal state of things. And you can see, you know, Microsoft and Sun are developing their own cloud things.

And IBM has some cloud tools, and you could see that they would have cloud things. And so the question becomes, if you have very popular clouds, and they're not using Linux, because they have their own sort of operating systems or things that they're worried about, then it's just a decrease. Once you consolidate down to these few cloud providers, then there's less chance for Linux to thrive. So the other, you know, sort of related to that, or, you know, kind of a sort of a counterintuitive thing, you know, not all these threats are necessarily consistent with each other, there's just possibilities of things that could happen. You could see that instead of having a consolidation of clouds, you could have like cloud distro madness, where there's like 300 different clouds that you could run in. And, you know, that's, it's both a good thing and a bad thing for the overall Linux community. Now, if more of those things are based on Linux, then it definitely is, it's just good for the spread, the sheer weight and numbers of Linux. But on the other hand, it's damn annoying if you're a QA, you know, a QA person or a support person or developers or anyone, and you just have to worry about more stuff. And I mean, this is sort of like a concern of SOA things is that there's so many different SOA solutions for this complicated thing. And if maybe if we could just have fewer, it would actually work out well for us. So there is a danger on the opposite end of consolidation, where you've got way too much stuff, and you've got to find that nice middle ground there. So the other issue is that, you know, as with any technology, like I was saying, virtualization might turn out to be a more important thing. So you could see that if cloud stuff ended up, whether it consolidated or it sort of collapses an interesting thing, and there was too much interest from the Linux world in it, you sort of put all of your eggs into one basket, then you would be kind of left with no, you know, you could be left just like with not, essentially, you would have put all of your effort into this, this sort of phenomenon, it would have not panned out so well. So you've got to make sure to diversify, essentially, is what I'm saying, sort of the interest that you have with Linux. And to be frank, this is, you know, something that, you know, you want to avoid happening,

not something that could really necessarily happen. But it is definitely, you know, because Linux obviously is, there's plenty, like we were saying, there's netbooks and there's plenty of physical servers and things behind the firewall. So I really doubt something like that would happen. But, you know, there are people, for example, who, who based, who based their sort of fortunes on SOA and, and as SOA sort of consolidates and fizzles out, you know, they got to find a new technology to base their stuff on. So you never want to put too much wood behind an arrow, if you will, in case that arrowhead is just porcelain or something. So having gone through that SWOT analysis there and the cloud thing, I think, you know, it's always nice to have some pretty immediate takeaways, if you will. And I think the conclusions that we can draw from this stuff is that in the worst case scenario, you know, I guess actually the worst case scenario would be putting all your eggs in one basket, as we just discussed. But the worst case scenario is that really, this is all just a bunch of chatter about cloud computing and nothing really changes. And things just kind of plot on the way they were. I mean, there might be more people hosting stuff and Amazon might be there and so forth and so on. But, you know, nothing much would, the dynamics of things would change very much. And in that scenario, I think the Linux world will be just fine. I mean, Linux has been around a long enough time and it's really cemented enough into the IT world that, you know, whether or not cloud computing is successful or not is not going to really determine, you know, the continued health of the Linux world, right? I mean, it could greatly, you know, it could make, it could boost the Linux world a lot more. But it's definitely, if cloud computing is just a bunch of crap, it's not really going to damage the Linux world all that much. Now, of course, the best case scenario is that you have more and more Linux instances because more, you know, the majority of cloud stuff out there is based on Linux. And then in that case, essentially, you know, instead of worrying about other operating systems for servers and sort of enterprise grade stuff running, most of it would be running, you know, the best case scenario on cloud computing grids and things like that. Most everything would be running on Linux and, you know, that clearly, you know, like, again, from the keynote yesterday, people always like quoting however many like crazy Wall Street transactions and everything that's running across their technology.

Because it really is a statement of the importance and the worth of that technology. And at the end of the day, if you're a technologist, that's kind of part of how you judge part of your own worth is like, here's what I've worked on. And here's my, in the positives, it's the word sort of legacy. And part of that legacy is the usage that I get from the software. So that really is the best case scenario that you have to look forward to is more and more people using Linux. Now, more, you know, like, I think if you want something interesting, you can think about, as far as like moving towards that best case scenario or doing something with cloud computing, what I spend a lot of time thinking about and worrying about is how if cloud computing becomes even moderately successful, we've got all this cemented, you know, legacy applications that are going to need to transition to a cloud computing based world. And the question is, you know, it's never really a question of how greenfield applications, brand new applications are going to fit into a technology because you just write it to fit into that technology. The question is, how do you move existing sort of brownfield, if you will, applications into the new technology so you can benefit from the new technology with all this old stuff that you have? And there's not really a lot of answers in the cloud computing world. And there's a little bit more in the virtualization world, just because virtualization, and I mean this only in the best way, sort of like a dumb approach to this. It's just virtualizing the whole stack instead of shifting the way the computing happens like in cloud computing. But there's not a lot of discussion and technologies and thinking about how, if I went to any given data center, any given organization or enterprise or whatever, and I looked at the existing applications in IT that they had, how would I move that to cloud computing? What's the technology that would drive that and the process that would do it? And I think anyone who's interested in cloud computing, that's something you can do like tomorrow, if you will. You can start thinking about solving that problem. That'll help you out, and it'll help out the industry and help out the concept of cloud computing. And with that, that's the quick overview of how this new technology, cloud computing, sort of relates to and affects the Linux world. So thank you. Thank you.

RedMonk TV · 🍕 Get more pizza.