The Cloud is Just Another Sun

Video in TIB AV-Portal: The Cloud is Just Another Sun

Formal Metadata

Title
The Cloud is Just Another Sun
Title of Series
Author
License
CC Attribution 2.0 Belgium:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Identifiers
Publisher
Release Date
2019
Language
English

Content Metadata

Subject Area
Abstract
You could reasonably call cloud services the crowning achievement in the world of Free and Open Source software. Linux and Free and Open Source software killed proprietary UNIXes in bare metal data centers and went on to dominate cloud services to such a point that it has even caused Microsoft to completely change their stance and embrace Linux and Free and Open Source software or risk the future of Azure and arguably the future of their company. Yet in many ways, this dominance has also bred complacence in the community. On top of all of those Linux instances are many proprietary services, abstraction layers and APIs that make cloud services easy to use for developers, but also turn them into the largest-scale proprietary operating system on the planet, where the network is the computer. Left unchecked, this proprietary operating system has the potential to undo the achievements Open Source software has made in the past two decades. The FOSS community has seen this "network is the computer" pattern before with Sun Microsystems and Solaris--a proprietary UNIX operating system that administrators ultimately loaded up with GNU software and free software services before deploying to the data center. Instead of Linux images running your dynamic Rails application or Docker container you ran CGIs in Apache and portable Java apps in Tomcat. Instead of disposable instances you had hot-swappable CPUs and RAM. Instead of S3 you had NFS. Expert users would use well-documented but proprietary CLI tools and libraries to interact with the OS and manage their free software processes. Yet in the end, administrators were subject to the roadmaps, whims, pricing structures, expensive hardware, and overall vendor lock-in from Sun. For all of Sun's talented engineers and sophisticated hardware and software, the freedom and values from Linux and Free and Open Source software combined with low-price commodity hardware ultimately dominated the server room. This keynote is part history lesson and part rallying cry. Proprietary OSes and services aren't dead, they just morphed into the cloud. By remembering why Linux was important in the age of Solaris, we can apply those lessons to cloud services before their proprietary APIs and vendor lock-in risk undoing the freedom, open standards, and overall progress our community has made over the last 20 years.
Loading...
Axiom of choice Greatest element Multiplication sign System administrator Floppy disk Set (mathematics) Information privacy Disk read-and-write head Sign (mathematics) Strategy game Different (Kate Ryan album) Negative number Software framework Damping Office suite Information security Physical system Point cloud Decision tree learning Software developer Stress (mechanics) Data storage device Cloud computing Bit Numeral (linguistics) Process (computing) Right angle Quicksort Information security Freeware Resultant Laptop Point (geometry) Server (computing) Windows Server Virtual machine Computer Depiction Number Twitter Goodness of fit Internet forum Internetworking Touch typing Operating system Computing platform Metropolitan area network Form (programming) Addition Default (computer science) Demo (music) Kälteerzeugung Content (media) Cartesian coordinate system Shareware Software Personal digital assistant Data center Point cloud Game theory Internet der Dinge Window Library (computing)
Group action Building Code Multiplication sign System administrator Source code Workstation <Musikinstrument> Sheaf (mathematics) 1 (number) Boom (sailing) Open set Mereology Roundness (object) Different (Kate Ryan album) Core dump Encryption Software framework Series (mathematics) Exception handling Physical system Mainframe computer Enterprise architecture Software developer Data storage device Bit Special unitary group Perturbation theory Flow separation Radical (chemistry) Befehlsprozessor Hard disk drive MiniDisc Right angle Smartphone Quicksort Freeware Spacetime Server (computing) Open source Virtual machine Software industry Computer Code Field (computer science) Product (business) Number Revision control Frequency Workload Goodness of fit Prototype Average Computer hardware Operating system Energy level Plug-in (computing) Computing platform Form (programming) Mobile Web Focus (optics) Standard deviation Demo (music) Projective plane Cartesian coordinate system Compiler Shareware Database normalization Bootstrap aggregating Software Integrated development environment Personal digital assistant Point cloud Codec Game theory
Group action Building Java applet Multiplication sign Workstation <Musikinstrument> 1 (number) Boom (sailing) Coma Berenices Disk read-and-write head Mereology Front and back ends Formal language Web 2.0 Direct numerical simulation Web service Virtual memory Different (Kate Ryan album) Oval File system Information security Physical system Point cloud Mainframe computer Area Collaborationism Software developer Data storage device Electronic mailing list Bit Cloud computing Special unitary group Computer Hand fan Web application Category of being Document management system Internet service provider Website Quicksort Middleware Server (computing) Sequel Virtual machine Mass Computer Binary file Rule of inference Number Pi Root Internetworking Computer hardware Business model Computing platform Mobile Web Domain name Default (computer science) Focus (optics) Standard deviation Dot product Inheritance (object-oriented programming) Database Cartesian coordinate system Human migration Database normalization Software Integrated development environment Personal digital assistant Logic Boom (sailing) Data center Point cloud Video game Window
Gateway (telecommunications) Randomization Code Multiplication sign Boom (sailing) Combinational logic Function (mathematics) Open set Stack (abstract data type) Subset Formal language Web 2.0 Web service Perpetual motion Single-precision floating-point format Descriptive statistics Physical system Point cloud Mainframe computer Predictability Scripting language Computer-generated imagery Structural load Software developer Electronic mailing list Special unitary group Cloud computing Instance (computer science) Flow separation Entire function Befehlsprozessor In-System-Programmierung Process (computing) Internet service provider output Website Right angle Quicksort Freeware Middleware Resultant Writing Point (geometry) Server (computing) Sequel Real number Web browser Twitter Revision control Goodness of fit Internetworking Computer hardware Operating system Data structure Computing platform Module (mathematics) Multiplication Standard deviation Patch (Unix) Server (computing) Interface (computing) Database Uniform resource locator Word Software Query language Data center Point cloud Communications protocol
Point (geometry) Server (computing) Game controller Code Multiplication sign Real number Workstation <Musikinstrument> Source code Combinational logic Open set Product (business) Different (Kate Ryan album) Internetworking Computer hardware Series (mathematics) Data compression Computing platform Physical system User interface Addition Software developer Structural load Keyboard shortcut Shared memory Special unitary group Equivalence relation Type theory Software Integrated development environment Data center Quicksort Figurate number Freeware
good afternoon thanks everybody for
coming so my name is Kyle Rankin I am the chief security officer for a company called Pierre ISM of we make privacy and security focused laptops that run free software and sunna phone but that's not what I'm here to talk about so but before I talk about this topic I thought it would be sort of important so I'm
gonna be covering some content in this talk that might upset some people and so I would like to warn ahead of time talk about some of the content that might be some people may find on settings so upsetting so if you are upset you know ahead of time you know you can you can leave ahead of time if that's going to upset you so if you are a veteran of the UNIX wars I'm going to be talking about the UNIX wars not in great detail but in some details so if you are a veteran of that war and this might bring back some sort of post-traumatic stress of remembering what it was like back then there will be depictions of some of the battles just letting you know I'm just wanted to put that out there okay so all of you now you know you know if you develop proprietary software just a heads up I'm going to be saying negative things about proprietary software so if you write free software on the side but your your day job is getting paid to write proprietary software I might say something negative and it may upset you hopefully not we're all friends here if you use the cloud because I wanted to make sure they include everyone so if you use the cloud I might say something in this talk that upsets you offends you make sure you question or be concerned about your choices or my choices too so just putting all of that out there so now that everyone's ready we're all safe if you feel uncomfortable we you know we can always leave them we can talk about it afterwards okay so all that other way I just wanted to start by talking about how we won I mean congratulations everyone I mean if you were to talked about talk to anyone about Linux right now with in free software what they would tell you is we won and all sorts of ways we're super excited about how we've won we tell people if they will listen to us about how much we've won and dominate the world so for example Linux the fauces everywhere if you were to look turn over a rock and it's probably running Linux or at least some free software somewhere right and everything is running Linux so your pocket probably is running Linux or maybe it's not but at least the dominant operating system right now if you dig past all of the proprietary code and get right to the bottom of it you'll find a little bit of Linux in there and a little bit of free software so and that's in a lot of people's pockets the majority of people's pockets in the world in your home right now while you're here when it's not ddossing other people's networks you have devices in your home that are running Linux probably your fridge your refrigerator might even be running Linux your toaster I mean people in the olden days would joke about met BSD running and everything including a toaster but I think a lot of the smart toasters these days are running Linux so it's ubiquitous in your company's code so if you are a developer who's writing software and you're starting off from scratch with something one of the first places you're going to look is something a place like github and look for free software libraries or frameworks or other things that you can use to incorporate and then write your proprietary software with that so it's very common I mean that's it's the main way that people do software development these days is look for existing free software and of course in the cloud so I mean Linux and free software just completely dominated the cloud if you look at cloud offerings today the majority of the virtual machines that are running in the cloud are running some form of Linux usually the default is some kind of Linux in in addition to that a lot of the services that they are offering a lot of the use people the way that people use the cloud is to run free software services of some kind or their in some cases their own applications that they launch if they are not free software they are at least often built using free software frameworks so we've just completely dominated the cloud I mean it's it's really when you ask most people about this they either talk about we won because of your pocket or we won because of the cloud but I mean so we've won with the cloud so much that Microsoft now loves Linux not only do they love Linux but they have professed their for Linux in public on numerous occasions some more romantic than others but in any case they love Linux now this is something that would have been unthinkable for a long time ago but it was largely a result of how Linux thoroughly dominated the cloud and caused Microsoft to look into their cloud strategy in if you're trying to sell a lot of companies into moving into your cloud platform which you view as the future of your company and they say okay great and they say we have this cloud platform is great we can run all all of your Windows servers that you currently have in the cloud on our cloud and they say ok cool but what about the other 90% of our servers how is your Linux story and so they had to dramatically rethink how they approach the cloud and now are huge advocates for free software and huge advocates for Linux they in fact they love it so if you're gonna point to another reason why we won that would be a pretty good reason because a lot of us who were originally who've been around in this community for a long time that was like the enemy you know like there was this since like forums would just it the em dollar sign stuff that I'm sure some of you probably still do like a lot of that stuff that was going on forever ok so but but if we if we won then who lost you know if there's winners and losers than who lost well I kind of already talked about and most if you were to ask somebody in the community that's been around for a while who lost wouldn't if Linux won then they would probably say Microsoft but honestly I don't like really we're mostly just talking about Windows server here because like I said they've completely it's it's sort of like there's not a lot of people in here probably that would say man IBM you know they'd be really mad about big blue from back in the 80s right because they started spray-painting touch graffiti in the 90s on sidewalks and stuff and every or maybe it's a 2000 anyway everyone's like well yeah they're a big advocate for Linux now and everyone is not concerned about anything in the past and so a lot of people still sort of dwell on Microsoft but the downside of doing this I mean I would caution you against doing this too much is if we continue to look in the rearview mirror about the battles we had in the 90s and the early aughts in the meantime we've had we've allowed all of these other giants to sort of come in and take their place and we don't notice because what yeah but remember that time when Microsoft's like embedded Internet Explorer wasn't that you know that was that was awful you know so I would suggest not to dwell too much in that who else who else lost a son so Linux is success was largely at the expense of son in their dominance in server rooms you could argue that in addition to Microsoft Windows server not getting as much of a footprint in the data center Linux also supplanted the existing unix's that were in the data center and son took the brunt of that ultimately being acquired by Oracle but we're gonna get more into that research in motion a long time ago everyone sort of thinks that every if you're a if you're winning today and someone's dominant today that it will always be that way and no one can ever you know kind of fall from their pedestal but everyone was using a phone from this company Research In Motion a long time ago that was the platform celebrity everyone from celebrities the dignitaries all the way down to your average person in an office we're using research emotion phones but their operating system and their company was really affected by linux's dominance on the cell phone market and in particularly providing this nice platform for treating applications on phones and proprietary UNIX often in many cases lost because of this in some cases that companies went out of business in other cases they're still around but had and in some cases proprietary unix will always be around probably there's still administrators running AIX systems in the number of places but largely linux success was that the expensive proprietary linux or unix also embedded OSS you know all those Internet of Things devices that we talked about before that are ddossing everyone all the time and never being updated that's because they're running Linux and they're not running a proprietary embedded OS there anymore usually I mean there's these companies are still around but on the average if you're going to make a new toaster that is smart and sends you a tweet whenever you're is done it's probably going to be running some sort of Linux embedded OS with some sort of proprietary application on top of it okay so so we won but but did we really win though maybe we won at some point but I get this sense that if we did win if you would argue that we did win then we're losing we're starting to lose some of that ground that we gained when we did win that battle so what do I mean so when I was a kid and had my first computer so to date myself it had floppy disks and the way that you would get software's in my case I would go down to the grocery store and they had this little cart that had different shareware floppy disks that you could buy and that's when I you know my first exposure to games like Commander Keen & Wolfenstein before it was 3d and things like that and the idea behind the software was you would get a floppy disks and it was essentially a demo and the idea was you would share this floppy
disk with your friends who would install it in and for free what you would get is a demo or like the first level of a game and if your and then if you wanted the rest of the game then you would pay money and you would get what we would now call the pro version of the game that would unlock all of the rest of the love of the levels well we have the same thing today basically if most of the applications on our smartphones or some form of shareware one way or the other it's either free and the money is being made by advertising or otherwise capturing and selling your data or it's or there's they it's free in a sort of like a demo where half of the features are disabled right and so and if you want everything enabled then you you buy the pro version and in all of these cases the codes not available it's all while there are free software applications on mobile devices for sure the gray majority of them are some form of shareware your company's code so I mentioned before how great it was that your company is using all of this free software when it's writing its internal applications Evette it then closes right well of course there's exceptions and hopefully there's a lot of exceptions in this audience but on the average if you were to go to a startup and who's really excited about all of these frameworks that are out there that allow them just to get bootstrap and running an application they would then take that application using all of those free software frameworks and their company's code would then be locked behind the walls of the company and and not be released so there's open for so this is a phenomenon where it's sort of dovetails a lot with startups so the idea behind open core is that you have some section of the code base that is free software and then surrounding it I'm in some way either via plugins or an Enterprise version everything there is a proprietary version of it where the source code is not released in it and that's usually how the company makes their money often open core companies will start so when you're doing this startup usually you get a very small amount of money to sort of seed the startup and you can't afford a lot of developers and so a lot of times an open port company is what will happen is you will create a free software project and put it out there and that way you can get community interest and it's almost like you know sort of demos and get some build some excitement about the product you're building get a lot of free development and then most most important when you're doing a software startup is to have some sort of prototype that's that works that would be suitable for enterprise so you can then provide that demo to venture capitalists and get your next round of funding so usually what happens then is you get this community going you get that first round demo you give it to you then can demonstrate that to venture capitalists who will give you you know a million or two million dollars or something or however much and then you can hire some of those open core developers who are in the community the best ones bring them in and then they can start working on your proprietary enterprise product while occasionally back porting things eventually you if you're successful you have this nice successful company that's starting to build revenue and then you can hire a lot of developers you get more investment and then ultimately when ends up happening is the the free software thing often becomes sort of a demo so it's considered off on that toy where the the real we wouldn't really use the free software for enterprises because it's kind of a toy like proprietary software is what you need for the enterprise and so that's and ultimately you have this weird separation where features get added to the enterprise version and the open-source version of the free software version the community tries the back port them which is weird because the company is supposed to be maintaining both but then the community has to back port these enterprise features in some cases things like encryption I would like my network to be encrypted well you must be an enterprise if you care about that right so the cloud so we won the cloud right there's there are more Linux VMs and free software in the cloud than anything else but what's where is the software that runs the cloud you know I mean we nominally think well yet it's probably all over any Linux and there's VMs that are running Linux but the way that most people use the cloud and I'm going to get into this more later is not very open there you wouldn't really know that it's running free software or that it's Linux you generally interact with it through a series of api's they that are based on some service that you don't have the source to and you're talking to it via that way so in many ways especially when you look at the cloud we are returning to the time before the rise of Linux and Foss there's a lot of this ground that we gained that we're losing for a number of different reasons that I'll get into so but then that's a so in essence it's the same sort of thing where we have a brand-new played away at the same play but we have brand-new actors so we have new tech giants but the battle is largely the same it's still sort of this notion of is free software a toy or is it legitimately you should be used for enterprise and is it good suitable for production all of us would probably say yeah well of course it is but some of that's being called in the question and if you listen to a lot of people's marketing who are selling proprietary software they have that sentiment where basically well yeah if you want to dabble if you're a hobbyist you go over here to the free software side but if you're serious if you're a business and in and you need to depend upon software then it has to be locked up in proprietary that's the only thing you can trust so part of the premise of this talk is to say well it seems like we're starting to rehash a battle what can we apply from the last time we had a battle like this in 1/2 the situation today and so the the bulk of this talk is going to be talking about that so let's talk about the unix wars' a little bit I told you I would so there's a period of time where the network was the computer this was you know before let's talk about before the dot-com boom because I'm going to get into that so the network was the computer so that means we had we had these end-user workstations that were not very powerful and in many different ways they lacked resources in the number of different ways and so what we would do is move storage because you know we didn't have huge hard drives where we could you know saw all kinds of things on our workstations we didn't have a lot of storage so we moved move storage computation and even desktop environments all on the network and so you would have this this is after just a durable like a dumb terminal that had just text we're talking like graphical desktops would be on the network and come over the network because it was the local works they should may not have enough resources to handle it this is a time where the transition from mainframes into large UNIX servers so instead of like one or one standalone gigantic mainframe you would start having smaller but still cheaper but still relatively expensive servers running Munich's there was a number of main hardware vendors and each of them had their own flavor of proprietary Unix that would run so son had Solaris IBM had AIX HP had HP us and then Silicon Graphics had Eric's and each of them would bundle that Unix with their operating system it wasn't necessarily something that was charge for but it was their operating system that would that would come with the hardware you bought the thing is this these operating systems were proprietary we couldn't see the code but they did have open api's and they had a POSIX standard which meant that you could write applications for one and be relatively assured that if you followed this standard your application would run well on another UNIX this was incredibly powerful for developers because you'd I mean as you can imagine you want this kind of compatibility what ultimately ended up happening though was people would then take the existing proprietary UNIX especially as as this became more dominant in some of the user space became a little stagnant and people who would put new tools on top a lot of so the GNU project started rewriting a lot of these user space tools on the command line in a compiler and having that all be free software and so one of the first things you would do as an administrator would be you install all of these good new tools in fact a lot of the first free software companies that that became successful became successful out of providing support for free software on proprietary platforms not on Linux the other thing to notice each of these companies have incredibly amazing engineering teams very very smart people really doing cutting-edge work in a lot of different fields it's not like all of them were doing we're making incredible advances in computing at the time some of the engineers from each of these companies moved on to do incredibly great things after they left that particular company the main focus was on hardware features in redundancy so what I mean by that is you would see things like hot swappable CPUs in RAM in one gigantic machine and then the software that could that could handle doing that without losing any workload on the machine there was also this sort of assumption while and I'm sure in some cases people that blends that there's this assumption that if you have if you were if a Sun shop you were a Sun shop and you only had Sun Hardware if you're an IBM shop you only had hardware from once vendor and that's where you would get everything so let's contrast that with
the cloud wars in the cloud where's the network is the computer there's this notion that we have these small devices that don't really have enough resources to do everything themselves and so they reach out to the cloud for resources so things like file storage because there's not a lot of storage often on mobile devices computation you will run all of that in the cloud that's in even desktops a lot of people will have remote desktops and there's a couple of companies that provide security solutions where they will take a desktop and sort of stay in boxing in a cloud environment you can use it if it gets infected you can you know wipe it and restart but it's all on the network so in this case the transition instead of from mainframes to large services from large servers that were local that turned into really small but remote virtual machines so in this case the players were talking about Amazon with AWS Google with GCP Microsoft with Azure in Oracle and there's a number of other providers that all want a piece of this cloud pie because they see that it's you know in many cases the future of providing like profitable services so again this is a these are services are all proprietary but they do publish their api's their api's aren't necessarily each other unlike with with Unix where you at least had a standard to code against generally with these providers their API is well open or also proprietary and there's not a lot of collaboration between making them agree with each other and like with Unix you would take this cloud environment with its with proprietary api's and then you would put Foss tools on top of it it could be Linux VMs it could be your fossa plication of some kind like MySQL or something some sort of rails application or your entire you know your entire environment again all of these companies hire amazing incredibly smart people and to build these platforms and they have all kinds of advanced features very smart people making these things and every year you hear about all of these new features all of these new things that make your life easier all of these new innovations on these platforms in this case the focus is more on software features than really hardware features I mean hardware features are saying well we added a new 7x large whatever you know like they they add maybe a faster machine but it's not really that's not where the innovation is the innovation is on all of the software services and the redundancy there's also the assumption that you get everything from a single bin these these vendors are not operating thinking well yeah you're gonna get a little bit for me and a little bit from Azure and a little bit from GCP or whatever they all assume you're going you're going to use them for everything
okay so let's go back in time again so now we're done with the we're still in the middle of the unix wars but let's scoot ahead to the dot-com boom so the dot-com boom was this era where if you took a business model and then put it on the internet you could get a lot of intricate 'el and so a lot of people did this they would say well what if we took a grocery store and we put it on the internet or what if we took a bookstore and we put it on the Internet and everyone's like that's a stupid idea or they said that's a genius idea I don't even know how to evaluate you because it's on the internet the rules don't apply anymore so if you if you knew that if you took a business and put it on the Internet that you could make millions and millions and billions of dollars then wouldn't you start to spawn new websites that were running some interesting business and so there's this huge pusher of new websites all over the place everybody needs a website now but where before they didn't know what the World Wide Web was now everybody needs a website your company needs a website everyone's trying to buy up all the dot-com domains because they're now are instantly super valuable lots of servers lots and lots of servers to make this work which means huge data center growth if you're in this if you're in Silicon Valley during the time they were building up taking properties and building in data centers they couldn't they couldn't build them fast enough they're often waiting lists to rack servers they were just filling up constantly all the way through the bust but they were I mean it was an amazing time for data center growth in the area and Sun Solaris dominates now I'm sure there are fans of other platforms I would say well mate I don't know like we didn't use that or whatever but a lot of people's data centers were running Sun and Solaris machines during the beginning of the dot-com boom so normally you would have Hardware maybe running nsca or Netscape web server depending on you know net crafts would say it would be NSEA mostly but you one of the other you would probably run some sort of C or Java middleware that would do some sort of you know application logic and then all of this would be housed in a really big can there go down Oracle database the one database you know if you if you have environments now where there's the one server that can't go down it's probably your Oracle database from this era so most of the software development for these comms were for this web these web applications were done on proprietary workstations at the time the proprietary workstations were probably running a Windows or Solaris so there's this sort of famous quote that son put out as part of their marketing saying we're the dot in com to sort of demonstrate their dominance with dot-com servers they were actually referring to a the fact that they were a DNS server at the time the dot is it's weird because the dots not even in com it's the missing dot the invisible dot that you don't see at the end because they were a root nameservers DNS nerds yeah all right I'm not a tattoo guy but if I could tattoo DMS I very anyway that's that's personal okay so let's go ahead to the next combo where what we did was we took a business model and we put it in the cloud which you might say well that sounds an awful lot like putting something on the internet and I say yeah that's exactly the say it is exactly the same thing so there's but this time it's a massive push for new web services because what was happening it was was everyone was starting to use this computer in their pocket as one of their primary computing devices and everyone saw this huge business opportunity and creating some sort of application that they could put in this computer in your pocket that would talk to this web service and somehow make your life easier and all of these businesses started moving all of their platforms to cloud providers to provide the computation and the heavy lifting you needed because you were running on this sort of underpowered computer in your pocket lots of the us lots and lots and lots of VMs you needed lots of VMs for one they were cheap and they were small and they were sort of underpowered unless you spend a lot of money on them and the other thing was they were notoriously unreliable but it's your fault when they went down because they told you we're unreliable it's not our fault that things are going to crash all the time the network's going to disconnect sometimes it's really just an unstable environment that's your fault so everyone started learning how to build very redundant systems which meant lots of hymns and being spread out all over the place and then you would do that and then there would be a massive outage somewhere and somehow your service will go down anyway and it would be your fault and then you would find out there's a bigger there's a bigger basket your eggs were thought you thought your eggs were in three baskets but they were actually just in one basket that was bigger so this also this spawned the huge growth in cloud infrastructure companies Amazon had the head the head start here and so they massively expand it became sort of the default but there are a lot of other platforms both just pure cloud infrastructure where you would get vm's and other ones that would sort of contain an application and run an application for you that started like growing and now there's a lot of consolidation in the industry but there's a lot of these companies that were spawning from this and in this case we're talking about a migration off of traditional data centers so a lot of people were also saying well it's about time to buy new servers but boy though they're expensive and the network engineers and system in there running them are pretty expensive too what if we put it all in the cloud we wouldn't need it nearly as big of a team we wouldn't have to buy this have this big capital expense that we have to make every three to five years and so a lot of companies started making this migration like I said AWS dominated it still dominates because it had this great head start and everyone just sort of the fact they'll started moving to it although again there's a lot of contenders now that are trying to win the crown of in the cloud wars and but what you normally would see is some sort of Linux VM it was running either Apache or nginx the middleware was a blend of a bunch of different languages probably really Python Java JavaScript and usually and then on the back end you would have like a MySQL or PostgreSQL sequel database and that would be that's sort of like the traditional stack early on at least before it became more sort of service-oriented again software development during the dot-com boom - 2.0 was on proprietary workstations at this time mostly on Mac books and then on Windows machines as well this also saw a lot of common Foss services so what would happen is a lot of people would say we like this no sequel database and they're like everyone's using this no sequel database and they're all creating their own no sequel database putting in in a VM and the cloud providers started looking at the most popular services that people were using say when what if we hosted that for them and they would just go through us and we'll make it a little bit cheaper then would cost to set it up themselves and they could just use our custom service our proprietary service instead of setting it up themselves and so they started doing this and then putting sort of an API in front of it to manage it and you started seeing this more and more and more where if the service is popular it's only a matter of time before it becomes a new custom service all right let's go back in time
again so we're back at middle of the dot-com boom and Linux is the is dismiss as a toy by unix's if you were to talk to any of them about well yeah I was thinking about maybe setting up a Linux server I've been hearing good things about it the all of these major unix companies would say it's a toy it's it's it's it may be it's good for a hobbyist but if you want to do real work you need a real production quality operating system you can't handle real load with this but then we have the release of the Apache web server and it's hard to underscore how important the Apache web server was for Linux and its success in the data center because what it did it had this in particular it had this great feature called virtual hosts and what that meant was you could post multiple websites on one web server where if you're spinning if you're buying a bunch of expensive individual servers and you can only run one website on them that's kind of a pain you can run multiples that's great it was cross-platform so you could run a Linux and Apache on commodity hardware you didn't have to buy expensive servers you could buy a random Intel server and run a website on it you also had a lamp stacks and this is something that replaced a traditional sort of Sun and Oracle blend that people would do and this is a combination of Linux and Apache MySQL is the database and Perl originally as this middleware and so there's this service called this standard called CGI the common gateway interface that would allow you to write snippets of code that you would then put on on a server that was running CGI if you hit a URL it would execute that code takes some inputs from the from the web query do some work and then output a result output HTML to the to the browser so ISPs started doing this thing where they would allow you to upload snippets of code ultimately PHP became more popular for this but you would upload snippets of Perl or PHP code to an ISPs web server because they wanted to offer web services and by doing that what they will let you do is you would execute it on the fly with someone hit this it was amazing mind-blowing if you would hit this you endpoint it would execute this code and give you this return there's this other problem that cropped up where people were buying is incredibly redundant but single Sun servers to run Oracle and they're very expensive because they could hot-swap CPUs they could ha swap ram they had all of these amazing features but they were expensive and it turns out some people said well what if I just put my SQL on three servers that were about five thousand dollars because their Intel and if one goes I don't have to hustle I can't how swap CPUs but I can take a server out and so people started doing this and this is starting to dramatically eating in to profits and it's starts to the trend that ultimately replaces proprietary Linux in my opinion with a proprietary Unix with Linux and so Linux Foss then when salinas wars so it's good
ahead to there's this new new technology that just came out that's referred to as serverless and what services is if you were to look at a cgi script which is a snippet of code that you could run somewhere upload and if you hit an endpoint it would execute server lists is that only its proprietary because back in CGI days you could take a CGI script and put it on your ISP you could put it in your own server running Apache with a CGI module or you could put it on a competitor it didn't really matter but with the modern the modern approach the CGI is to have sort of a custom a custom language a custom API that only were ends on one cloud provider but the same sort of story otherwise the thing that the interesting thing about server lists is that you have this further separation between the developer and the actual Foss services and Linux that are there because you're not really interacting with a nikkie Linux server that's running Foss services you're just sort of dealing with an API the ultimate goal is honestly just working with api's not to have to deal with servers or not have to deal with services just sort of right api's this also conveniently makes it hard to leave the platform because if your entire code base isn't even transferable because it's using all of these little snippets of code that use a custom API it's that much more difficult to leave is is there Linux or Foss behind these serverless service services I don't know I haven't seen the code so it's possible but I haven't seen the code so who knows so all that to say this goes back to sort of the premise of this talk which is the cloud is the largest proprietary any cloud infrastructure is the largely proprietary OS on the planet so AWS Google Cloud if you were a lot of people don't think of them that way because I think yet but it's split up and a bunch of these little VMs and be like well what about mainframes don't didn't mainframes have a bunch of these little isolated processes but they had a single operating system in the cloud any cloud infrastructure largely operates as one single unified but proprietary operating system where's where's the code if you wanted to run an instance of AWS or GCP or as you're at home could you do that could you download and build their code and then run a smaller instance of it yourself you know you couldn't but we don't really think of it that way largely because the api's are open right we think what would the api's are exposed so that's all that really matters because they document their a well of course they're gonna document their API is that's otherwise they're not very useful but a lot of people mistake documented API is with openness so usually like today's des if you were to have a developer or a system and work on the cloud today as its intended to be used and they haven't didn't have a lot of background they would get very minimal exposure to Linux and Foss generally speaking them I'm my prediction in the next five or ten years the job description for both will become the same which is basically I'm writing API glue but I might be running for different services but the jobs will sort of converge into someone who's basically writing API glue code that you the cloud the other interesting thing is a lot of people just sort of accept vendor lock-in these days it's odd because if I say there's a point in time where vendor lock-in was like a bad word and you said they're like oh yeah I don't want that and now a lot of people are saying well what's the big deal with vendor lock-in of courtly you would be foolish to switch to another platform because my clouds too big to fail it will never have any flaws there's never been any history of any cloud having an outage and Michael in my provider will always stay in business you know but here's another question okay fine maybe it's too big to fail but what happens if your cloud wins or what happens if a cloud that you're not using wins and yours loses why do you think the cloud provider would do then how you think their innovation and all of this rapid development would go in their pricing structure would go if they won and they were the dominant player or put it another way if they won would protocols on the internet really need to stay open if all of the websites on the internet mostly and all of the web if all the websites were running on one cloud provider wouldn't they would they need to communicate with each other using open standards couldn't you foresee a day where someone says well yeah but all of this HTTP stuff is pretty slow we're going to replace it with with a custom service so if you're inside our cloud you get this all of these benefits of our speed if you're outside the cloud of course you're slow and awful but if you're inside the cloud you get all of the stuff it's not that hard to conceive of the other thing is if all of these services replaced Linux pm's with something proprietary would anyone notice I mean those of us who go to the cloud and spawn a Linux VM would notice but people using the cloud as its intended to be used would have no idea you could just swish out are my no sequel service with one that's running proprietary code instead of a standard free software version and no one would notice and a lot of people wouldn't care so what do we do like us what do we do about this well the first thing is to
look back into why did the Linux and Faust wind to begin with so I would argue that Suns dominance early on led to some complacency in some arrogance the whole thing of what Linux is just a toy it will never do any and that also led them to sort of dismiss some of the advances that was making and innovation started slowing it but not so much in the hardware but in particular for the user interface so things like you would commonly hear people talk about I would get my workstation open I log into my son's ster server a type of Camaro I made a mistake backspace oh why does the backspace work and this is like a small little trolli comment except that there's this notion that will ya the backspace doesn't work you're not using a sudden workstation if you use over certain workstation with a Sun keyboard it would work fine all you have to do is add this sty command and it's totally fine but what it points to is more of this arrogance that went along with will we're never going to update thar to have modern command-line arguments which nowadays did you know that you don't even have to specify like what compression algorithm to use when you're when you're compressing something it just sort of figures that as sort of amazing but but they didn't they didn't do that like back then it was just like it was just straight up vanilla tar vanilla everything that's why people install the new tools to begin with because the free software world was innovating faster here's the thing there were more Foss engineers in son engineers a lot of people like to think well yeah they're but they're super smart engineers are the best and that may be but there's also only so many of them that are working on software and when you compare to the world of free software developers there's a dramatic difference Linux and free software innovations outpaced the rest of the innovations that everyone was making that's why everyone shifted to it in addition because the the openness of everything to cross-platform compatibility it turned Hardware into this commodity you could run this platform on anything ultimately and even made sons look into doing x86 and eventually even opening their their software in in Solaris but by then it was too late ok but that doesn't tell us
what we do right so so what do we do what one is you should recognize proprietary code when you see it or I should say when you don't see it there's a lot of times where we just had this we don't even really think about it we a lot of a lot of people just focus on is the API open but ask yourself when you're using a service ok well that's cool if I wanted to run it myself could I download the code can I at least in spec the code even if I don't know how to run it myself to see what it's doing large a lot of times the answer is no I would say we should resist proprietary services vendors make it very easy to only use them for everything that's the whole point vendors have always done this I mean the vendor lock-in has existed forever I started a series on this at Linux Journal to talk about some of the old stories of interlaken because i'm I'm honestly not sure that a lot of people are familiar with the problems with interlaken so that's the next thing ideally everyone could do who's been around and understands what can happen when a vendor has tight control of your entire environment is tells you to tell your stories share with people some of the problems with that here's the thing is the fact that we as a community are open and we focus on compatibility and we and we share our source code is our advantage and we don't always realize that as an advantage but it allows us to do certain things that proprietary companies would never do they could they couldn't do because it would impact their it would impact their business too much to free some of their software it allows us to take it we can take advantage of that and allows us to do things that they can't please share your innovations if you're coming up with if you developing software that solves a real problem please release an under a free license I know it's very tempting to take it and then use it as the start of your new startup very tempting to do that but please consider instead sharing that innovation with everyone else so that we can get a look we can all together get further along than we could otherwise and finally sort of want to pose this question to everyone so one of the big things that that drove Linux is success in the data center was this lamp stack this notion that there was this combination of all free software services on a free software operating system that could run on commodity hardware that was not just it was easy to set up and it was powerful and it was suitable for production loads what what's the equivalent of that today what's the what's the equivalent thing that would cause everyone to question their use of the proprietary services that they're using today that they're relying on today is there a way I could could one be created or is it that if someone were some enough to come up with something like lamp today they would be too tempted by the desire to close it all up and turn it into a start-up and we wouldn't see it hopefully the answer is some of us I mean some of us are already thinking about this problem and are working on some solutions to that but you can talk to me about that later but hopefully some of you are also thinking about that what is what the next things that we can work on to re-establish free software as the the de-facto way that we do things on the internet because it was that way at one point we all grew up some of us grew up appreciating the benefits but it's not a given and if we don't do something it won't always be there thank you [Applause] [Applause] you
Loading...
Feedback

Timings

  684 ms - page object

Version

AV-Portal 3.20.2 (36f6df173ce4850b467c9cb7af359cf1cdaed247)
hidden