Merken

Contain yourselves

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
so high and it starts Howell to build systems that scale how we build in the past how the features looking like and have a scale that of this a lot of false info on but of despair that help around as you can see it tugboat flying trying trying so hard to prevent the catastrophies or the helicopter using so 1st little bit about me and I'm address I've been doing i t now for what about 2 decades or so and hang out a lot of the CCC I've been to nearly every 1st come so far thanks very much for having it's always great to be here but In the past I've been very privileged to have a small Amsterdam start up called putting that comes when we grew that 1 each year 60 per cent year-over-year so we start out with a few machines and people 5 people in the office of and we ended up being the world's largest travel but inside I think by now 10 thousand people and billions of revenue each year and it's this talk is a little bit based on what we experience there what where we've seen what works and what was the analysis very much hyped in the in the newspapers or the Seattle weekly magazine or anything in the area or if you put hackers news that's it's not all accurate in my opinion so that's why the targets contained yourself so I think before you follow the since I did the beginning of step that uh out of it from daily operations on nowadays to a lot of information security and financial industry I which is very good because it allows you to better the deflection unit ball day-to-day anymore so you get a good overview golden that so yeah I've seen that is that it was still works and what's what I would consider a waste of time so to them let's let's start with me the good old days I don't know that have a look back in time but we all know the good old days everything was made from wood and it was so much better right you can see that it's too if you will follow some people on Twitter they go like in know I am not up all the citation anymore it's it's horrible the people or will technology is getting worse and worse I think I'm I'm SUSY concerning woodworking unfortunately along the the old days they went on really they and I would say they mostly sucked if you look at the corporate IT you had processes t was always the gatekeeper ities job was to prevent people from doing work this the old joke from Gilbert but he was a model of modern active preventive information I t was seen as a cost center and you had chargebacks had all these processes which in the end of the day to get people from actually getting work done and it was it was horrible the there were so many silos you had to uh the the typical life cycle for an application and the company would be you have the architects the design the software from you have developers write the software there will be consultants installed the whole stack and afterwards handed over to the operational team to run it and the users to use it but knew that she just hated because it's so bad the but as get fixed yes of course but Circuit found 1st then they get fixed so usually this 6 months at a time between releases because it's so complicated every release take down website for 2 days or something because you need to reinstall everything on and if you like you get the bug fixed in time because it's QA review the change advisory boards needs to agree that this is important it needs to be tested again and in the test was not correct so it's section of something to happen this year and will be next year as I said it's really your will it also kind of like i t done by checklist a checklist monkeys go to the chapter still get bananas at the end of the day not a good time so since then good things
got better they got the thing was the Internet and the Internet is really really great you know all the song that this has been a tremendous increase speed because the replicate that applications you have code that native software providers they want they don't want brick-and-mortar style they only have a website they can do upgrades everyday stuff gets approved and delivered to the customer immediately 1 fibers later the world has seen the future you don't need to test it anymore that long you don't need to ship about you don't need to buy the customers with upgrades and you see Microsoft following rapidly when 10 uh policies were almost the updated automatically and you see software like sales force as a software-as-a-service application being very successful because they can of the features so much faster than oracle apps for any of the other SAP as if he comes to mind or other European systems so this is it seems to be a real success so everybody wants that you have all companies brick-and-mortar stores enterprises they see so that applications of sites and they like this which is really great we want that to but they can't compete they have this six-month cycles complicated but have the processes and what sort of comes along there's much quicker they can deploy quicker they can do upgrades quicker they like new features this is a new article and the news on your height they can pull it out and say week or something they have something which works customers flocked to it because it might not be perfect but it works and and so much quicker with Oracle I need to wait 6 more months of something and maybe I don't get it or how it works but I need to buy another 1 and 2 that the concept behind the disruption the Web startups you about sort of faster and more agile leader the new old companies and they eat their lunch yet is a much-touted example it's not really a very ethical company but they think very successful with the with getting the message across that they are much better than the current taxi industry but it's quite important to to see how was it done right and let's do a quickly track how the native companies can be that cricket 1st of all they don't generally need a very very clear is subjective what is it that if you want to solve L for looking for example it was a very simple idea the idea was let's make it very easy for customers to book hotels worldwide that fits into 1 sentence that's all you need to know because at that point you people can define it can decide that OK this is what we want to do we have ice in my area of responsibility I see these 3 4 issues which prevent people from doing that simply because this too much confirmation yeah this too much paperwork to them can I find a way around can fix these things with such a simple message everybody can visit takes also a canon ownership of these issues and get them out of the way to make the booking experience easy so this is the 1 part but you also need a very clear and tactical objectives that that could be allow Crick role of of the features we want to make sure that a developer is able to roll out a new software feature in arises from what saved typical example could be we have a project owner who understand the business part of the application here's an idea let's say it would be really cool if you could try this out he finds developed in the morning was available maybe it's use 1 1 so and he has 1 and he talks of development of others as you know the the public take me 3 hours so by lunch developers starts and he's done by 4 clock in the afternoon would be nice if we could just try it out and see if it works so the idea the developer should be able to what about without any interaction say in 30 minutes tops that's a good feature will attract tactical go in the beginning
we had this it worked fine and development this slowed over the wall start production operations and the other alternatives it works on my machine that typical example from systems operation to that would be very rich and images possible copying e-mail off you know because the data center not really that good it's what I said before the development rights operations once it's a very difficult to scale out at that point because you will have to much friction this again the tickets between teams and all that the inside of this will have to go to scale out l development needs to go faster we can't wait for operations to do these things to the 1 hour install take on the website for 4 hours that will that will not work so development got a lot of interesting inequality processes or different changes in processes uh this agile programming this this extreme programming pair programming and then poisoned this several in the industry for the development process the idea behind all those find a way to do make it clear what the priorities for development of have them right the software without any further distracted from customer changing requirements for example that you will have because things it would be really nice of that color could be red and tomorrow now I thought about it it's blues really nice of maybe mold can get me maybe that's not a good way because you're you are hunting of the for that situation is actually really good because you defining the requirement and then you have 2 weeks to implement and afterwards instead what time the custom can say yeah that's OK or now I would really like the green and the effect of that actually we can to the green uh which is that this is that important does need to happen right now or so we put it on the back of the backlog because remember if we do it right now that means the next 2 weeks we're doing and a very of customers at that point we decided what it's fine with the red said before it's it's more important that we get a new features installed there was quite well but but once the code has been written it is to out to the service because you web company and you want to allow of the features again right that's attacking code quickly as so this is automated FIL deploys typical example we want to automate the code deployments this allows for the fast and easy to processes before um you look into different options you you might start small with this span a copy of so you just copy the code to website we start the over all is good if you're working a bit larger so you don't have 5 observers have high and that takes a bit longer might want paralyze used key is a sure PCP made you looking at RCA because the data is getting larger so depending on the needs that different options like you need to find the right ones for yourself but what about those people about the operation people have been lazy that right they should be easier as well that an automated machine configuration changes this not for example that the shift this sensible as 1 of the greatest once a sold media so you see vengeance and or potentially you might have written your own shows not the wisest decision because you need to do something supporting all of sudden but when they're just the vertices a configuration management system written as a joke called just themselves with this so get out it's it's if you going to show it's an isolated fix it otherwise I would recommend real Chef Puppet terms salt and some of those of the user once you also looking at alternating machines selection because it really doesn't scale well if you have to send someone out with a constant because the disk to the data center and the like this goes in install this goes out next machine that's not good you will be looking at booting of remote installations maybe using ILO whatever you have to ask what if you're lucky and you using Amazon while service API contestants on machine maybe you work with golden management you want to install by squared to really up to you and you can completely new so I wouldn't want personally I'm not a big fan of the domain images 1 because I find it lacks a that the agility because if I want to update 1 packages so I would need to be spent in image that will that everyone reinstall everything if I use a regular say puppet for my of its management just the target added 13 is 8 and and I don't need to install the machines and good but it's up to what he works for you of the benefit of all these things that you get a repeatable and transfer configuration change applied it doesn't matter if it's a single machine or hundred thousand things although it doesn't matter if some them at all on the cloud change Alexi Dr. makes a difference in all and community should be done on the 5 minutes there are companies who have if there were 2 experimental they know exactly the capacity planning they have every quarter they buy a new heart of 400 thousand they know how much the need this quarter and if you need a machine it's already running the wrecks just connect to other people might be designed know what we can live with the high latency for scale-out because HP delivers takes a five-week so if you know we need something 6 weeks that might work for you to know if that's not working you can't afford the ending spare hardware around the look at Klout look at virtual digital ocean hexamer might be something decent and this again the following but thousands and thousands of options is all up to you the idea behind all this is called infrastructures code itself turned informed by exceed those who have been the solution lies in the eBay for hand crafting things and they've been very very successful in pushing the message how they work and they've been very active at conference like velocity of money to run generally I would suggest it's interesting talk with them if you ever need anything engineers and the top layer is only know what they're doing it's always very interesting to get new ideas are the people working on and laughter you to go to go right you probably want to test it has this is not a joke that's what the lot of people don't like this where you going to test it you to tested in a special testing environment how do you make sure that testing of mind is actually accurate to new production how do you make sure that the notes that you get a good understanding how you notice behaving if just below massive change SQL there can I tell how that will be behaving I would like you to get an exact copy of my environment that we need to get an exact copy of the behavior of the users I might be able to so if some traffic replay and online testing environment that I can make you spend a lot of time trying to recreate the production environment but generally I will always forgot something and my testing and not be as accurate or anything so a lot of problems of Giselle you know what we're
just going to test and production we're going to build our education in such a way that we can survive if a server fails during test and the Council will not notice they have decided they have found a way to some would booking was that every capacity tests introduction and meant that you would take 1 cell which is a set of from uh web servers and the database and the number of database servers they would just increase the priority in the local so they get more and more and more and more traffic and 1 point you see that response times go up so the point OK and with that we have found the capacity seeding of that cell and we have 100 of those cells so we can take the number of repressed could serve this 1 cell multiplied by a hundred and we know what we can do at that point you know if say all this is the peak you can then decide you want to August we are here and this is what we'll see in August the new safe or the and might need to talk to mention that the that we now spend time on fixing some performance issues would just find hardware really obvious attitude who knows again in the regions of so the point that the production and get more and more realistic about and this is risky absolutely it has a certain amount of risk I would not recommend that for example for with the metal manufacturing company probably not the best idea ever now I would not recommend to test out the latest code to write on the productions with the full workshop plants and the bank maybe not on the many past but if you work in PR side or landing page for my research and what not might be good enough again it all depends on the situation and if you take all the states in the added together you get this 0 my god damn lots and to make all this work to satisfy the business objectives you know you can't continue the old style and you know mutable silos they don't scale there horrible barriers to entry they they make and I will basically and this that so the integrated teams where you developers and your operations people and they need to work together they know each other they they talk they sit to the sink on they have full responsibility for the whole stack you can't just building all the dimensions of their operations that the US was then doesn't work and we need to get rid of that the style away had the built 1 deployed or what I explained earlier with the the consultancy architects all that needs to go that needs to be wonderful T and if you do that that way you can actually deliver the velocity agility what but you would say normally this is a common sense you know it's it's really important to really good that developers and Europe engineers talk to each other but this is it's actually very rare in a lot of companies that it's aware that this is a joke says consensus so where it ought to be a superpower and and then what was the name has been codified all these and the people interact with each other have full of shared responsibility have transference that 1 team that supports all 2nd transform and that's the idea of develops and it's not a tool thing especially how highly interact with the people it means failure needs to be accepted because if you working that fast yes you will fail from time that needs to be will and so that that will happen a booking for some we had a budget to each quarter was 100 thousand euros or a half a million whatever was we lose this year through human failure and that's that's just the costing business and that's the pump price we are told you willing to pay because it allows us to be so much faster than the the competition and the budget for failure for example something completely new in the airline industry for the non-coding parts it's just not done and when they heard about the things that we have a section the interesting thing I'm not sure we could get that into our culture it's it's a new day full support from your management team to make that work but if it works it into be a very critical of the typical success stories so this would be at sea booking of common Facebook Google Amazon Netflix all these very large very successful companies to develop things in 1 way or another but there's also small 1 start ups because of competition if you talk to them they actually explain how that works as very much the same thing so you know that that's the words but if you look at it from a fascinating observation but because if you look at the the way people as and devised what works for them where all the bonuses and how do we judge their work you'll see that development lots while they are in the same team they still have very different incentive for you feature development is generally purely results it doesn't matter how they got to the code they got it doesn't matter if they had to take shortcuts and the code beautiful that really doesn't matter the important part of features that what might be that what that releases 5 % cricket which is really good because it saves us from having to buy 10 more machines this year it will be great it doesn't matter that the Haskell Curry is not the nicest all that they use copy and paste from somewhere for your operations however the it's a very different game the the result is not actually that important it doesn't really matter if the deployed was error-free often have to be rolled back or anything it was a compromise was not no no effect the important part is that it's well understood how old they got from the old release from you really is that this is repeatable that they can do not just 1 machine that can all machines that they can roll it back that they can verified wasn't successful is running is women well was not running well these are all things which are important flops but not for development of and prescribed form a colleague of mine from looking at different extended very nice to see this talk that I go away or shows that the patient is myself where he says the infrastructure development operations is just for the worst case while the future development development as judged by the best case for feature government is always all this is really also has a nasal feature for chronic development as you can see in the next government that they never ask it is a really cool feature which really nice how does it work in the best case the scandal examples always so what's the worst case how much worse would get if we apply that patch where abouts how bad the crash that's a not necessarily mean what the most rewarding experience you've just wouldn't new page his super proud and they're like yes so of great that's it I assume it's not for everyone but the doesn't make we have in amazing tools for the future is totally awesome but because nowadays we have container they solve all
these things and specifically we have talk about which is the best ever this then rolled from red head to there's local work on Doc getting secure conspired folded isn't and he always event and its stock under a buck I need to say 3 times the rate really well and as I said we have we have this amazing new technology call containers the most famous example would be a doctor and they promise to solve all the examples all the problems are just as quot before they can cancel 1 tool it's amazing and because all your code you don't need to worry about implications of machines anymore only nicer contained in the application domain you loaded on a boat and the boat goes ghosts and that's boat goes to production and that's and stopped and and you don't need to worry about dependencies begin deployed anywhere because every continous saying it doesn't matter if you're running in an Apache and on the Linux machine on a Windows machine whatever so all in there amazing and this part just works that's most important so if we look closer Would but actually does we can have a look at the website that's what because says straight to the website it's accurate portrayal part of it is but were somewhat about reading it as I highlighted the important parts of the it's a platform for developing answers and to assemble applications from components and then the get deployed production as fast as possible so if we have not it actually not that and if you look at the right it doesn't always synergies that don't talk about web-scale it doesn't have real mobility whatever so it's it's kind of this it additional description I would say that there are problems with this for example here the it's developers and so both cold out of sync with it and he talks about getting killed and deployed production so wait that's not what is at most that's what developers to it's actually that that may yet another develops a pack up code into production have we seen that before and I'm not talking about the the old joke about the right ones 1 anywhere from John I'm not thinking in we find evidence of strong enough it's also that the me like they if he is binary blob which in my machine goes on and on production and fear that we've seen that before we decided in the past that that's not the way we want to work it does not work could I had somebody can govern comedy would have that attitude but it brings the old joke to might as well the if you don't understand package management you forced to be implemented badly because that's exactly what doctors being public the post at all I have is that undergo shouldn't production you work and I'm a bit surprised actually received the technology behind doctor is it's not the best but they actually manage quite nicely to get together to the masses and get containers ready for use yet all there seem to be interested in this practice often many dependencies that something which have been solved 20 years ago 15 years ago was at get RBM has all the young ones bath and say but if all of the dialog handling of lots of areas belonging to this yes but that's where it fails like the right and the development of an but the dependency is itself a pattern that's well understood and and this is especially if you within a company is a little bit better so it's a bit different on the wider Internet for example when you have these nowadays these appliances VMware appliances so if downloaded test like this piece is offered at that point architecture makes alot sense but for the warning out applications in enterprise and things like that so we kept the what we have a tool which contains loss of best practices we from the past on building scalable infrastructure but we're not doing the the work anymore we're having a development team which is packed in software in somewhere you know nobody knows exactly how works because as a stock of file but that a national guarantee that would getting the doctor file is actually what's described in bill filed by doctor is working on finding things but again that's just more like yeah it's really coming from the person who who claims that's coming from there is not a guarantee that would be getting there is actually the accurate of some of you get binary blob applied on top of another binary block if you look at the examples you'll find something like let's take this cloud imagine we'll add this class image binary from winter to we let this Apache binary on top and we'll take the mice fell binding and talk and real energy will convert files and we got a red and a web server from is there's no guarantee that if I build back the same doctor file in 2 months again that I'm getting the same because maybe you leave you that data from this the binaries what you've seen in there that in the doctor files where code is downloaded and installed to the parties that form the baby built this no guarantee that what be getting is actually recreated will anything the that too but on the other hand if you're building your own you probably for that because you can make sure that you people that immediate occasion these you collecting is applied company stands but let's say if you take the basis of 1 to energy you almost like the safe but I must feel safe enough with lots of and that the Pentagon company sites and some the interesting thing is that we we kind have had that before that you get a new binary blobs applied top binding blobs and ownership production to 1 application of we had about 40 years ago it died 25 years ago and was called small talk and that's a good meal because shipping system state and timing has never really been really manageable and has been a good idea because small talk tools were often written in a way that he is the code but you know we just a life patch the patch that and that's what you want and the ability to be example in company for example is imagine you have a part of talk 0 you have the nicest confinement of thousands she's very heterogeneous anyone OK how many machines would Hôpital Mozelle also stopped batch do we actually have if you're a old-style enterprise you can ask you cigarette and 1st of light so that they know and you can fix of genes usually happen to this factor it knows the answer to that if you are a chef or high-intensity ontology and running and so will solve this on the left to right all managed if you understand the answer affordance on there's no reproducibility and there is no way to tell what's in the package which becomes a bit of a problem if you need to decide OK it's time to all those packages all those images or not you can fix that by installing property know it just for example but less obvious 1 because some use of implication containing when you actually believe full virtual machine inductive and that point winding Borland what kind of just packed up the codon continued destination for example but the good news is that
even if it's even if it would be horrible right it can't be that forward conservation that others and that's actually something thing I find doctor the technology is OK there use cases for comments Dr. successfully managed to to get them to get a better interface place on the Linux Containers and see if we look at for example the set of people they have been bootstrapping go I think into you want to and 1 5 from now and they did with darker containers because that was the best way how to get a complete built environment separate from your post change and then build something cross so they can monitor on arms 700 is what they're working on that that's actually quite cool because it's quite hard to change routes because you will always pull information for more information from you post you could do with full motion machine but at that point you have a lot of hassle so that her contains was actually choice and that's or if you are say you're a keyword hosting provider like you might decide that it's not a nice idea to have a lot of the web server where we have 50 the service of 50 host 50 users on the separate access to the new status you PHP for example you could just decide every that's ever gets sick every customer gets older continuous just mentioned PDE in there and the PHP dealing with PHP fast PHP or just a piece we model or whatever the 1 1 1 movie more passenger what Python will 1 1 and we'll have than the Nobel so all the proxy server from Dr. then we direct the traffic to the right machines essentially really really interesting use case but I don't see a lot of people talking about that it's always still that's suppose our approach production of that but if you have a seat it's working as quite nicely separated but if you don't you can get out of that and it's it's it's kind of like giving it to someone on the internet if you if you take a 3rd party when the common but if you build the community itself you would get more security for example because you money you make sure that the Apaches running as nobody in there or the user would be getting way more separation than if you want multiuser it should be on on the same machine so it's not perfect it's not him but is way better than what we have currently in a multiuser hosting environment for example and so that's why I say it's a very sad thing because all the hypothesis concerning container as just an impact format and I it's it's very backwards because we we want the about benefits as we've seen day-to-day work the everybody wants the life I working financial industry I see a lot of banks considering how can we use those how can we use the lessons from develops and all that in our environment we went to the local cricket we we've been during the OK was taken website for 4 hours in the morning for clock and role of the new release every quarter and have 50 people so get up early to test and everything is OK but confident about and I know that certain banks for online G animals is very very the head of the quality modern computing seen and they they go to the developed days they try to get a lot of input from the community how they can do things better and some things work for them something horribly on just they're trying to get a 20 you know 30 year old stacks of code sometimes 40 all still 1 in Cobol into modern environment with the 1st which which has never Council which never consider those cases when they when they build these technologies but for some things that actually were so that's quite impressive but yet with the Devils we wanted them benefits but that's completely ignore everything you know that has a winning move that's the it's not really call progress I think it's called backwards completely backwards and that's 1 idea and I'm not a fan of the current hype of the the gathered container where they're better ways to do that if you look at
for example the i scalability right so yeah that's the way that he would be really easy scalability how to do it only that is also very interesting to read how other people this fascinating means how Facebook scales out how Google scales an the so very interesting take away the current high I think there would be microservices and protesting management kinetics it's the latest greatest it will solve all the scalability because what you need to do is you take applications you separated into special services so for something about what right you will have 1 service which doesn't uh occurring you warehouse if you have still stuff you have 1 different service which will the authentication will have 1 service which will do the billing and if you get a nice little amount of services that way you'll have a nice benefit that you can take down service for maintenance while people can continue serving 3 aside maybe can't have you people looking for the 5 continuing the rebuilt or something but you full what will still be working and you can see this is Facebook does that for example very often but that you reload the page was suddenly this story missing which was there before when you check boxes loading of its slowing anything these also different servers and back and if a service is not working that's only find you can still use Facebook just missing is small feature things so the it is very nice and it if we can package those up and containers we tend to have a whole fleet of land on on a few machines or anything and I mean the management environment but the problem with that is that faces with this works and where you can get benefits from that but realistically chancellor it's not company it's project of the this is the saying division example would have is you have your community quality had container side said before you manage your hardware you manage it really is a puppet show for and if universal machines on Amazon uh my you probably use on them to because he wants all you want to you don't wanna install your central so is all the time all over again I think if you're not using the ideas from on so you will have some kind of competition consists of an idea getting containers but it doesn't come with 1 and not this suggests that you don't get 1 for that so you're looking at what do we do not always communities right that's possibility is that and I spending time to build up a completely new infrastructure just to mention contain without getting any of the synergy effects and then we and from the existing conference that's just not a ring of of fluid and this is a different talk of mine which describes a bit more how we've actually scale-out method is being best lessons learned from beginning but 1 thing we learned there was those takes time it does not happen overnight you will not have the situation that today we had 5 visitors today but tomorrow will have 5 million suddenly consistently over the next year so we need to be able to look to scale out from 1 day from 5 1 machine to 5 million to 500 genes that just over half of them with every automated you will also from 1 to 10 from 10 to 100 to 102 thousand to 10 thousand 200 thousand students there you compute the need to revisit the whole architecture nothing will working and examples 1 machine you that they have a tendency to probably kick-started but didn't do anything else 100 machines you would probably be looking very much about the public or other confirmation systems thousand machines you will definitely need something automated to install and configure movement from 10 thousand machines yeah happened will not work for you anymore because it's just so much changes everything and there are so many failures something which deals better takes up machines which fail out of bones and that point a very funny code from a Facebook engineer was so I'm looking at a machine which has a narrow at developers as machines acting funny I log into as as is age doesn't ask me for my key so this means I have a lot of the machinery for all these years was spoken before so that just trashed completely without even looking at it anymore because yeah I would normally not limited I would not in front of and it was but it would have that's something you will not that attitude you would not see was 10 machines attend machines you would still be looking at what's actually going on I need to try to understand what's going on how see how to fix it subsequently different things and at that point when you need to be lined up against completely and that he can decide is it now so so for me to go for containers school for example the absolute and the evidence corresponds application that works if you're workshop or bank something was 3 4 5 8 maybe 10 applications the hassle of running a separate course management system for that's always so look into the your existing tools of can reuse them and the other 1 is also a business perspective do you really think that it's a good way to spend your time now develop all the configuration infrastructure and management of research all the code framework for Europe also contain the situation or anything on the thing it might be a better way to spend your time in writing the features which attracts customers to you because customers like features they couldn't care less how you're running you know if you're running in container a few running bare-metal if you want a virtual box so whatever it is you like that OK so and I think that point machine at the kind of the end of last time have to more interesting means 1 is
also 1 it's called the future it's a future Paul bigger it's on it's a very it's a cat view of the current state of discrete systems 16 back into that go every day it's you rolling on the floor laughing at interface how a business guys coming to that developer and asking so they say you you know how the new things were created in Spain to me how works and he wants a and you want a single that perhaps something very simple and he ends up with about 80 to 90 containers probably on 18 vertices or anything in a bit of management resources but it's really great and scales up very nicely for assessing whether Hillary history and the other 1 is if you more interested in how we didn't how we build looking at that point is a role of surveillance condition and again so slight the extensively the business requirements a company had at that point what they decide was important how they decide how this that and what they have to accept for a company to to be able to do it of debate how much there is to be expected expected they how much effort to put capacity planning and what errors will you find on the way what the traps what works well what doesn't work so again very interesting read the and other note if you should be in interested in mice trail or it might have performance and the analysis of next month's there's a quite interesting shield Pocono like that the Poconos Performance Consulting the company they have a yearly conference call proposal like this year it's events of September this a very nice how but he was built and is such an amazing technologies as bare-metal hypervisors and it's different technologies such as Apache Mies frail and Pearl and I billions of money over that period and uh I think the genotype is boring . com but it's study because the words and that I guess I'll think about freedom him and questions politics of that but I expected regret that that that that this is all about the idea of the increments if you have any other questions we don't ICU and things always happen to share anecdotes and wisdom state I think
Offene Menge
Web Site
Bit
Prozess <Physik>
Mathematisierung
Adressraum
Supersymmetrie
EDV-Beratung
Kartesische Koordinaten
Whiteboard
Computeranimation
Virtuelle Maschine
Freeware
Informationsmodellierung
Prozess <Informatik>
Software
Konstante
Vorlesung/Konferenz
Hacker
Softwareentwickler
Ereignishorizont
Hilfesystem
Analysis
Softwaretest
Nichtlinearer Operator
Videospiel
Computersicherheit
Güte der Anpassung
Einheitskugel
Physikalisches System
Checkliste
Programmfehler
Kartesische Abgeschlossenheit
Software
Twitter <Softwareplattform>
Rechter Winkel
Dreiecksfreier Graph
Digitaltechnik
Garbentheorie
Information
Bit
Prozess <Physik>
Punkt
Virtualisierung
Reibungskraft
Wärmeübergang
Kartesische Koordinaten
Service provider
Eins
Internetworking
Rechenzentrum
Datenmanagement
Skalierbarkeit
RPC
Code
Trennschärfe <Statistik>
Vorlesung/Konferenz
Urbild <Mathematik>
E-Mail
Verschiebungsoperator
Softwaretest
Internetworking
Nichtlinearer Operator
App <Programm>
Zentrische Streckung
Hardware
Güte der Anpassung
Ruhmasse
Strömungsrichtung
Biprodukt
Entscheidungstheorie
Konfiguration <Informatik>
Konfigurationsverwaltung
Dienst <Informatik>
Forcing
Rechter Winkel
Digitalisierer
Wiener-Hopf-Gleichung
Projektive Ebene
Extreme programming
Schlüsselverwaltung
Programmierumgebung
Message-Passing
Geschwindigkeit
Web Site
Subtraktion
Mathematisierung
Automatische Handlungsplanung
Interaktives Fernsehen
Term
Code
Virtuelle Maschine
Knotenmenge
Domain-Name
Benutzerbeteiligung
Ungleichung
Fächer <Mathematik>
Software
Reelle Zahl
Mini-Disc
Endogene Variable
Äußere Algebra eines Moduls
Luenberger-Beobachter
Installation <Informatik>
Optimierung
Speicher <Informatik>
Softwareentwickler
Rippen <Informatik>
Konfigurationsraum
Bildgebendes Verfahren
Soundverarbeitung
Booten
Kanalkapazität
Gibbs-Verteilung
Physikalisches System
Quick-Sort
Objekt <Kategorie>
Flächeninhalt
Hypermedia
Dreiecksfreier Graph
Mereologie
Gotcha <Informatik>
Kantenfärbung
Unternehmensarchitektur
Streuungsdiagramm
Orakel <Informatik>
Einfügungsdämpfung
Wärmeübergang
Extrempunkt
Gerichteter Graph
Homepage
Deskriptive Statistik
Code
Mustersprache
Vorlesung/Konferenz
Feuchteleitung
Softwaretest
Business Object
Managementinformationssystem
Hardware
Gruppe <Mathematik>
Kategorie <Mathematik>
Mobiles Internet
Gebäude <Mathematik>
Güte der Anpassung
Ruhmasse
Biprodukt
Bitrate
Ereignishorizont
Entscheidungstheorie
Rechter Winkel
Grundsätze ordnungsmäßiger Datenverarbeitung
Server
Ultraviolett-Photoelektronenspektroskopie
Geschwindigkeit
Facebook
Klasse <Mathematik>
Systemzusammenbruch
Systemplattform
Virtuelle Maschine
Bildschirmmaske
Domain-Name
Spieltheorie
Reelle Zahl
Endogene Variable
Soundverarbeitung
Videospiel
Varianz
Elektronische Publikation
Menge
Patch <Software>
Bildschirmmaske
Streuungsdiagramm
Resultante
Bit
Punkt
Statistische Schlussweise
Kartesische Koordinaten
Euler-Winkel
Binärcode
Synchronisierung
Eins
Internetworking
Metropolitan area network
Datenmanagement
Skalierbarkeit
Bildschirmfenster
Zentrische Streckung
Schnelltaste
Nichtlinearer Operator
Softwareentwickler
Assembler
Datenhaltung
Quarkconfinement
Stellenring
Systemaufruf
p-Block
Dialekt
Teilbarkeit
FLOPS <Informatik>
Landing Page
Garbentheorie
Aggregatzustand
Web Site
Hausdorff-Dimension
EDV-Beratung
Zahlenbereich
Zellularer Automat
Code
Data Mining
Benutzerbeteiligung
Software
Zusammenhängender Graph
Response-Zeit
Softwareentwickler
Bildgebendes Verfahren
Schreib-Lese-Kopf
Ontologie <Wissensverarbeitung>
Division
Kanalkapazität
Physikalisches System
Energiedichte
Zustandsdichte
Flächeninhalt
Mereologie
Basisvektor
Computerarchitektur
Stapelverarbeitung
Unternehmensarchitektur
Bit
Punkt
Hyperbolischer Differentialoperator
t-Test
Kinematik
Kartesische Koordinaten
Euler-Winkel
Computerunterstütztes Verfahren
Statistische Hypothese
Service provider
Internetworking
Homepage
Metropolitan area network
Skalierbarkeit
Datenmanagement
Vorlesung/Konferenz
Auswahlaxiom
Schnittstelle
Softwaretest
Zentrische Streckung
Teilnehmerrechensystem
Hardware
Computersicherheit
Gebäude <Mathematik>
Ein-Ausgabe
Biprodukt
Softwarewartung
Arithmetisches Mittel
Dienst <Informatik>
Menge
Rechter Winkel
Server
Dateiformat
Projektive Ebene
Information
Schlüsselverwaltung
Programmierumgebung
Proxy Server
Subtraktion
Web Site
Facebook
Quader
Fluid
Mathematisierung
Keller <Informatik>
Framework <Informatik>
Code
Division
Data Mining
Virtuelle Maschine
Benutzerbeteiligung
Unterring
Arithmetische Folge
Perspektive
Fächer <Mathematik>
Softwareentwickler
Konfigurationsraum
Schreib-Lese-Kopf
Soundverarbeitung
Trennungsaxiom
Videospiel
Routing
Physikalisches System
Turing-Maschine
Authentifikation
Computerarchitektur
Energieerhaltung
Offene Menge
Bit
Subtraktion
Punkt
Gemeinsamer Speicher
Automatische Handlungsplanung
EDV-Beratung
Diskrete Gruppe
Computeranimation
Freeware
Knotenmenge
Datenmanagement
COM
Vorlesung/Konferenz
Softwareentwickler
Analysis
Schnittstelle
Beobachtungsstudie
Zentrische Streckung
Sichtenkonzept
Kanalkapazität
Systemaufruf
Strömungsrichtung
Physikalisches System
Frequenz
Ereignishorizont
Software
Konditionszahl
Computerunterstützte Übersetzung
Aggregatzustand
Fehlermeldung

Metadaten

Formale Metadaten

Titel Contain yourselves
Serientitel FrOSCon 2015
Teil 76
Anzahl der Teile 80
Autor Thienemann, Andreas
Lizenz CC-Namensnennung 3.0 Unported:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/19548
Herausgeber Free and Open Source software Conference (FrOSCon) e.V.
Erscheinungsjahr 2015
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract In this talk, I'll offer my contrarian view on the latest hyped containerization technologies, why I think the new approach is flawed and how we can usefully blend the old and new prevailing wisdom to build systems that actually scale. Andreas Thienemann

Ähnliche Filme

Loading...