An Expert Talks Containers and Cloud Computing

hi I'm James McGuire and our topic today is container technology which is uh certainly influencing software development and cloud computing and microservices and all sorts of other Tech topics to talk about that we have a we have a leading expert in the field with me is Gordon half technology evangelist for Red Hat hey Gordon how are you doing today hey James good to talk to you again so I and and I love your actual job title is technology evangelist I've always love that you have the word evangelist in your job title it seems like a wonderfully open-ended job title to have well and I went I used to be I used to go by Cloud evangelist and then that was too narrow so I went even broader with technology so I I I can pass myself off as an expert on absolutely anything even I was an industry analyst is a pretty good transition right okay great so you know containers I mean it's really the last few years obviously containers have been very much on on the upswing and I I think there's still a lot of confusion about virtual machines versus containers and how those two fit together and you know whether whether or not containers are are good for legacy apps and the kind of effect they have so let's do a little sort of a 101 background thing I think it might be basic for some folks but I think it's worth doing talk about what exactly containers are as they relate to Virtual machines so what what are containers and and why are they so ideal for for cloud computing sure it's containers have actually been around for a long time in some form they actually go back to BSD jails um think 1990s May 1980s maybe even and so they've been around for a long time basically what a container is and it simplest and containers become interesting for a lot of other reasons which I'm sure we'll get into here but in its very basic form uh Unix a lot of other modern operating systems have this idea of resource groups or process groups ways to let an application execute a bunch of bits to do something and it provides various types of isolation mechanisms so that those don't get in each other's way so you know one application running a system doesn't suck up all the CPU and memory and so forth so our other art are interfere have security holes relative to other applications that kind of thing so basically what containers were originally what BSD jails were originally were they added some isolation to those to those uh basically units concept those Unix processes and what has that happened over time was there were more Resource Management put on more um more types of isolation techniques um this came into Linux something called lxc originally uh and there were some actually some commercial products based in containers but they uh were they were always very Niche uh mostly service providers used them ASW soft was an example one container-based technology um and these changes what they basically did from the point of view of the user point of view of the application was they looked somewhat like separate systems so if you were running in a container you thought that you had all of the resources of a full physical system but in fact you only had some of them and in this respect is it's conceptually similar to Virtual machines except that you're in the case of virtual machines you have a layer basically sitting on the hardware and in the case of containers that layer is in the operating system basically and that's why containers are so tied up with Linux and the operating system well and of course for the virtual machine it carries its own operating system or or or the necessary operating system with it as it goes correct correct and in fact again not to spend too much time in the historical context here but I would argue the re there was somewhat of a battle between virtual machines and containers back around 2201 or thereabouts over who was which technology was going to win to split up physical servers these very underutilized physical servers do bombbomb had happened do com crash had happened no one had any money uh servers were often 15% utilized and suddenly there was these these software techniques that say Hey you know you can maybe run do three times as much work without buying any more servers and people were sort of going well sign me up using using virtualization you mean well use it well at the time using virt you could have done it using virtualization virtuoso with SW soft was arguing you could do it with containers sun was arguing you could do it with containers with Solaris um but what really happened was first of all it has an appetite for bringing in so much technology at one time so it was a situation where somebody was going to really win this and what happened was at the time uh there were a lot of what we would today call snowflake servers so by that what I mean is you've got an application on a specially tuned particular version of an operating system and those kind of had to stay together and containers really weren't very good for that because containers virtualize the operating system uh everything needs to pretty much be running on the same version of the operating system uh also we really a good match with windows at the time so the hardware virtualization idea which VMware initially went out at was just a better model for these data centers with all kinds of different operating system versions and applications and certainly the Enterprises flocked to you know virtualization in vmore in particular just I mean they built an Empire over you know by you know 2009 2010 12 somewhere in there it was like it was just VMware was it was long by the default uh default virtualization layer for the Enterprise yeah I mean certainly since that time um uh open source virtualization kbm in particular has made pretty significant inroads but you're absolutely right and partly it was increased utilization the other thing that happened over time was that virtualization uh got into things like live migration and storage pooling and all kinds of by the fact that you were abstracting the hardware by the fact that you were kind of inserting this software layer above the hardware the hypervisor the hypervisor the hypervisor you know in a way the first step toward softwar defined infrastructure in a sense it let you do a bunch of other things that were were really hard when everything was locked in on a physical server so um so so yes the hardware virtualization uh really became c not Universal but certainly became very very widely used in Enterprise as but it was the spring of I think it was in 2014 or 15 you know a Docker just exploded on the scene I think it would have been 2014 I'm not sure I can't remember the year exactly but suddenly it was like you know containers came out of nowhere apparently I know they had the the the the history to them but they they appeared out of nowhere seemingly and and they became this very rapid upward Trend in the Enterprise yeah absolutely and not just well the air prize and and probably initially even more so the web scale companies in Silicon Valley and the like uh but yeah and I mean I'm a little embarrassed to admit that I wrote a book called Computing next uh in about two 2013 2014 sort of where cloud computing is go where it came from where it's going and so forth and I basically didn't mention containers I mean you know you put a revised version of that that's what what well I I would have every just every well I did just I did just come out with first version of a new book uh that talks about containers and packaging and so forth what what is the title of your new book Ordon well the title is just in pre-release right now but it's from uh parts and vets to programs and apps and it's how software learned to package itself and basically taking these lessons from retail and packaging historically and there's a really great segue into what you just asked me uh and how we brought that into software and that was really what happened with containers in about 2013 is yes the containers took advantage of um this kind of historical operating system virtualization model but what changed was they really became a packaging mechanism a way to of application portability uh across a r of different systems including public Cloud so they so they they they really uh kind of took historical containers as a b you know as a working base but they they really added this patching mechanism and then well over time these orchestr orchestration mechanisms came in and really this very broad Cloud native like kubernetes from the orchestration layer and in fact one of the things that's um that's fascinating about what's going on in the container space today is that if you look at for instance um the cloud native Computing Foundation within the lyrics Foundation actually publishes what they call the cloud native landcape and it's this enormous PDF document I mean they could make a Pooler out of this thing uh is just all these companies playing in this space and it's really this remarkable in many ways you mean the container space you don't mean the orchestration space you mean well it's all part I consider all part of the cloud native space generally because the reason containers is in are interesting is again not because of just containers and something like um uh you know the runtime uh the runtime container format the image format standardized under the open container initiatives that's interesting but to really use containers effectively as part of really a a Computing system and a Computing system these days really means your entire Data Center and maybe multiple data SS takes all of this orchestration logging monitoring uh and all these different uh types of different projects different applications many of which are open source and they all play together with each other and that's really what's giving this again mostly open source ecosystem its power well well now when you you're talking about playing well together is is containers going to put is containers going to make virtual machines up are they going to live live together side by side happily well there's a couple of things going on first of all nothing nothing that has become as dominant as virtual machines is ever going to die overnight and there's still an enormous interest in you know I I don't really like using this word it's a little pejorative but Legacy virtualization but but certainly the it optimization side of things the look I've got I've got at least supp portion of my application infrastructure it runs fine it's not worth rewriting as microservices containers you know not worth moving to a public cloud and it really makes sense to uh to to continue to maintain and even expand uh that that part of the it footprint and you know Red Hat has a virtual has a kbm based virtualization uh product and you know that continues to do very very well for us just had new release of it so there's still a lot of interest of that uh in it shops the second thing is is that um in many cases containers and virtual machines get used together so classic you know classic example is Amazon web services um you running cont know they you ec2 is virtual machine based and you can abely run containers on top of that virtualized infrastructure you can run containers on an openstack uh in Virtual infrastructure on premise and there's also work being done with containers specifically within the open within the open stack Project work but you can also just have open stack and run a container platform like say open shift on top of that so I think the answer is traditional virtualization stays around the two get used in conjunction with each other certainly people are experimenting with running heavier weight you know fat containers as well and we'll see how this plays out over time but but certainly it's not a case of containers taking out virtual machines anytime soon yeah well good all right clear my misconception for me that I I've been somewhat confused about is that when I was sort of learning about containers I thought the containers meant that if if an app was wrapped in a container it could travel freely through many different Computing environments but I'm not so sure that's true anymore so I learn more about it I mean does it does it does a container need to be in a specific environment to be happy or can it can one container travel between various Computing environments so a contain so so first of all container containers are with within an operating system they're within Linux so there's a very tight connection between containers and predominantly Linux Linux operating system because it's the operating system that really creates that container environment containers are portable I mean that's one of the one of the whole ideas behind them that you have this portable file format you have this portable runtime can package up an application and move it from place to place now having said that um you start talking about complex a application environments you know talk about Google no you're not going to pick up Google and move it somewhere else probably not you have all you know you have all the monitoring infrastructure the uh the orchestration infrastructure the you know all the other the the connections to data sources um where physical locality May well matter so so absolutely uh containers provide portability but but as with any of these things as you get into bigger and more complex environments it's not just as easy as pushing a move button and expecting that it you can move it halfway all your data and all your applications Halfway Around the World to another data center gotcha okay so so a container is is not going to bounce happily from a Linux environment to to Windows environment like truly heterogeneous environment it wants well and again contain I mean containers are and earlier question about V you know VMS well yes it may make sense to uh to have your Linux containers running in a VM on you know an Azure or whatever for example um so and I mean there is definitely work going on in terms of kind of broader container interoperability um like like Microsoft is doing some work in terms of you know running Linux containers on Microsoft Windows for example and so the absolutely the goal is to is to increase uh an enhanced portability across as many environments as you can but but you're being a bit disingenuous as you just say oh sure you can just move it anywhere you want however you want so really the the the upswing and the trend upward in containers is is a victory for open source in many ways well yeah I mean again you look at uh you you look at the projects involved in again Cloud native Computing generally uh in terms of the uh of the oci runtime the oci image format from kind of the base level point of view but then you look uh across the uh Cloud native Computing Foundation which you know doesn't that's not where everything lives but there sort of a center of gravity and you know kubernetes fluent D Linker D core DNS uh Prometheus uh and lots of other projects that you know I'm not naming again there I think there's literally dozens of them just in that one Matrix and and I think it absolutely is a victory for open source and I think at the um sort of most simplistic level well yeah all this stuff is open source and it's enabling Cloud native Computing so yeah open source is know has uh gotten a huge amount of the of the Mind share and is delivering a huge amount of the badfit to users in that space But I I would argue actually that the sort of the W open source winning there's there's even there's more maybe even more important and somewhat more subtle points there because one of the things that this Cloud native Space is really showing is how powerful open source is as a development and collaboration model because yeah before open source there is no way I mean there probably aren't enough lawyers on the planet to write contracts that would have been needed to have collaborating in this way and instead Engineers from different companies just go up into to GitHub and you yes there's still lawyers involved for trademarks and and and Foundation governance and those kind of things but but I I think what probably was not was certainly not widely recognized and I'm actually not sure if any of the even any of the principles who were involved in early open source early on really fully Gro this kind of collaboration it would enable because I think it would I think it was it was much more wrapped in the user freedoms and those kind of things which are absolutely absolutely important but we're I don't think it was probably appreciated that outside of the IDE you know sort of the ideological and protection of the users I don't think it was widely appreciated just how powerful it open source would become as a development model oh I mean absolutely and and this this certainly is probably a separate six-h hour topic but it's fascinating how open source has sort of moved from sort of this developer you know very developer Centric world of the 90s where we loved it we get to all work together and then then like corporations slowly understood oh open source that can be a huge tool for us and I mean the war is over and open source has won and it it has gone so far from those early days of developer Centric yeah now you know I will you know I will say and I say in my in my new book that it you know it's not all you know unicorns and Roses and Rainbows and the like I mean on the other hand you know you do have these big cloud providers some of which are better at participating than others I mean you know give Google credit you know they they you know kubernetes came out of their internal work um and they have sarney on board now as a community manager for kubernetes and they've really uh I think they've really learned how to work better in the open um Microsoft has has made very significant strides obviously yeah Amazon perhaps somewhat less so um you're being very diplomatic Gordon very you know now you know a Amazon I mean Amazon does a lot and certainly they make you know they they've been uh you know they've been a big Boon for a lot of developers who walk up and running quickly right um but they haven't been as big a contributor to open source as some of the other equally large Cloud prop I think that that is clearly agreed upon yeah why why um and you may have mentioned this but I'd love to put it in a nutshell if we can why is why are containers so such a good fit for for cloud computing sure um well the portability we talked about right um the other thing I'm I'm not so it s depends how we're defining Cloud for this answer so let me answer a slightly different way why are they so such a great fit for cloud native architectures okay and couple of reasons so first of all the snowflake uh customized operating system images that we associate with virtualization we don't do that any longer um we're using very standardized uh operating environments uh very large scale out operating environments uh where the idea is the immutable instances you put them out there and if something goes wrong or if something UNT changes you kill it and start a new one we we're also of this idea and this is still I'd say a little bit in flux but this idea that we're tending to move from monolithic big monolithic application architectures to these more Loosely coupled service microservice based ones right CLE yeah so you have a small service a container where it's a mutable youve got lots of them you fire them up they may not live very long oh and a container is a small lightweight container right that can be fired up very quickly destroyed very quickly doesn't take up a lot of resources so oh that fits together and oh by the way we're also moving to these more uh more agile devops style uh development processes where you have a lot of automation automated testing automated deployments Canary deployments blue green deployment again very good fit for a container so so from my perspective uh what's happened is you have you know some sometimes things have been around for a while but you know the soil's sort of Rocky and Stony and isn't very isn't very good mix for them and then you suddenly have this you have this different type of environment with different types of priorities because the other thing I'd say is that you ask most um cios today you know you go back you know certainly around 2008 and go you know what's your number one priority saving money right uh you asked your CIO today average CIO day and they still care about saving money but they also really care about you know things like enabling my company's dig digal transformation Journey partnering with the business deploying new Revenue producing Services more quickly competitive Advantage right competitive Advantage technology is a competitive advantage and again this idea you give these sort of new application architectures to support new business models to support uh new types of services to to support new business opportunities and you can do it with these kind of very scalable agile sorts of infrastructure with containers uh developed using devops and all this stuff really kind of plays off each other and fits together so I mean you you've said a lot a lot of good stuff a lot of good information I guess I I I'm really curious about where it's going because it's very container technology feels like it's very much in flux you know things coming and going constantly where are we going over the next couple years with with container orchestration engines and with containers I mean where's this headed in the next few years yeah well obviously I wasn't very good at at predicting the last time I tried this well I tell you no one I mean no one could put a technology book in in print and then look smart three years later I there's no one who's ever done that it just doesn't work that way yeah well so I mean there's there's a few few things I'd say um first of all I actually think the technology is going to uh stay in a lot flux with these projects really there's still a huge amount of innovation going on what we're seeing is limited standardization where you need that standardization in order to innovate on top of it so for example um I I referred to the open container initiative uh before which again is also part of the laks foundation and in contrast to cncf they have a very narrow focus on what are some of the specific areas in containers which happens to be the um the image format in the runtime now they're also looking at distribution and potentially some other things um where okay we have something that works here today this isn't we we'd really like to kind of standardize in this area so that people can just plug into this but can have some confidence this isn't going to be rapidly changed now the chall now what I just said probably terrifies some of your listeners because want this idea that all this stuff is going be in flux and I'm gonna have to learn new stuff all the time and what we've actually seen Happening Here is particularly a lot of the bigger tech companies you Banks and so forth you know kind of in round one they go wow all this open source all this stuff I can just download and try and then you know a year later it's well my devs have really goodlook resumes now but I don't actually have something I can use but they Lear the devs have learned a lot and they've had a great time yeah but so so one of the things that we see see at Red Hat happening is um is not not do too much advertising here but we have a container platform open shift where we basically pull a lot of this stuff together and you make some choices about where things are mature do integration work that kind of thing and there's definitely a lot of interest out there in uh we are need know a de platform or we need just container orchestration platform for our Ops teams uh but you know we'd really like someone to do this for us and there's there's a there's a lot of interest there including do it yourselfers who have kind of come you know who we had spoken with and who were sort of like yeah you know thanks but you know there's we we we we're we're smart we can do this ourselves and it reminds me of the old days when uh when U you know there were Banks who had directors of konel engineering politics well what year would this have been that was probably I was say Illumina at the time so this would probably it's would been late 90s late 90s an actual bank had something called a director of Linux Kel engineering yeah I've been told that anyway I'm not sure I ever met actually excuse me this would have been early it would have been early 2000 yeah okay yeah that's fascinating there were still a lot of um there were still a lot of financial organizations they still do do a lot of customization but increasingly I think there's been a recognition that you know there is like an infinite amount of Technology work to be done uh out lot of companies and certainly at financial institutions and so forth and if some of that work can be easily done with someone who frankly is a lot more expertise in it than you do why not so your your your bottom line though is that the flux will continue going forward we're not going to get say 2019 it's not going to all settle down with a standard and just like continue on smoothly I mean I think we have standardization some are so you know kubernetes we've talked about for example right kuber you I'm not about to say kubernetes is the orchestration winner and it's going somewhat depend on workloads and that kind of thing but if you look at Community size you look Community momentum I think you look and I I feel reasonably confident saying that hey kubernetes that's going to be the mainstream step standard container orchestration PL okay now that's a that's a good right I like that's a good solid prediction you know and I think you can look at some other particular uh projects um in the cncf and elsewhere um you know I mentioned Prometheus there there really is uh again I don't think you're going to have a single monitoring alerting solution but that's something there seems to be a lot of momentum behind and you can kind continue down the list and I you know I think there's there's going to be areas where you start to say yeah that's that's going to be the 80% solution uh you know and then I think you look at other areas like serverless Computing for example which um are pretty new um you know Amazon announced Lambda I guess a couple of AWS reinvents ago function of I'm not sure how much momentum it has at this point but it feels like it is a lot I I'm sensing there there's certainly there's certainly interest in the idea and then there's the question of well what do you do with that that you don't do you know which you can't just do with a you know a regular container okay and set up your own event triggers and things like that and then you get into the whole question of you know do I really want write my application so it can only run on Amazon and it depend how how much do you want to go in that but on the other hand there's some open- Source serverless Computing I I actually really hate serverless Computing ter so functions functions as a service they like function as a service better than you like I like I like function as a service yeah so so there's there's definitely some work going on there and I think that's an air that's an area that we could see some expansion in so I mean I think you said it Gord any any any last thought before we wrap up today no I mean the I've been in the industry for a little while and I mean this has really been last few years I mean um have really been pretty you know remarkable I mean I think certainly starring about 2010 with Cloud going from well when we say cloud what we actually mean is a self-service portal on our virtualization solution to really genuine Cloud native you know clouds yeah also people going from you know the puet cloud is insecure no it's not the public cloud is insecure no it's not you know these kind more secure than inhouse it seems like that's and in some and in some cases it is well and my point is people are under are better understanding use cases and having more sophisticated discussions about them so so that's um but and then so so that and then containers coming in and I think I want to say changing everything we knew about infrastructures and writing applications but the whole devops uh Cloud native microservices containerized platforms Etc has changed things pretty radically um over the over over the last few years absolutely it has gone it has gone crazy the last say two to three years in particular yeah yeah and uh you know kind what comes next um you know I I think absolutely that you know I think a LW of innovation does does take place here but I think the normal cycle is you would expect some degree of stabilization in this space and you know there's certainly I think there's certainly two still questions over kind of how does the private versus public Cloud play out uh you know how um you know to what degree people do end up switching over to software as a service for certain things and again it's going depend in the size of the organization and so forth um you know I think you know certainly there's a lot going on and you know what you um you might like might you might call the SZ in developer space there's a lot of effort going on in in the development space you how do we make it easier for more people to uh devel for these business users to develop applications so I think that's something that is going to go on but I really just seeing this just uh sort of Cambrian explosion I me we didn't even get into things like I mean we've really been talking infrastructure here mostly so you know not even talk about the number of uh development languages and so forth out there exploding right um you know I you know iot uh is going to impact all of this uh that's that's in very early stages right it is um and you know there's there's still a lot of questions there but uh certainly companies like GE are getting enormous benefit out of it yeah g g is hip again it seems like with the iot thing is going on G G is you know it's pretty cool yeah yeah so you know so I mean those are I think sub of the areas and then obviously in applications you know where does the whole you know AI machine learning Etc type thing and all of this stuff isn't related to Containers directly but I think it does point it kind of is because things like data and application pability are important there um thing and it and it just points to there all this stuff runs on computers it runs on operating systems right you know so this stuff isn't going you know isn't going away it may be under the covers as far as the Afric developer is concern but is Appley still there and it's hard not to see they're going to be a continuing insatiable appetite for well you blanked out there at the end Appetite For What for computing oh absolutely no it's only get larger and larger yeah uh Gordon I think you said it uh thank you very much expertise it's just a lot of fantastic stuff uh it's unusual things ahead to be sure interesting things so thanks a lot I'll talk to you soon

This transcript was generated automatically from the video's captions and may contain errors.