Bestand wählen
Merken

You have a Cloud, now What?

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
good thanks thanks for coming but that's the almost last thought on this and this thanks for sticking around an up and coming to to listen to this and you have a cloud now what so long in I
used to to start working on cloud and really what I was looking at is how to get a
clone of Amazon E C 2 some working on that but she got 2nd a commuter on class and a member of the Kinsey so we know we worry about a low level details of actually how you create a cost that you manage storage network and sort really what I'm interested in personally is to get a clone of these OK because you know we've learned everything from Amazon and well without from Amazon we have no clue how they're building that kind of get at the problem using then but but you know we don't know much in terms of storage in terms of network we don't know how they're and creating all their services so scout black magic with what we know is that they have a bunch of services so that storage services of course they have Institute for Computer they have a archive service is another coming up with a bunch of analytics services of course Elastic MapReduce that came came out in 2009 marking is is now for a stream processing are redshift for a data warehouse so when you look at this in terms of clouds and it's nice that we all talking about hypervisor choice and performance and as the and and storage solution and staff works on this but that doesn't work is this version of the birth and things like this but you know when we look at Amazon really they're just spinning at new services left and right every every month every quarter we have new services and to me that's what you know it's about OK it's about building those services and then using them where there is for application deployment of or application the application development in terms of to infrastructure or actually having a petition and putting it on the yeah so that the landscape you know we see today in and yesterday the landscape in terms of Open Source AI Infrastructure-as-a-Service solution it's really are CloudStack OpenStack OpenNebula you get over it the energy we heard a talk by the guys from seeing info I don't think there was any by you characters but you get it this was really the 1st software that came out there was trying to our to emulate Amazon and create a clone of of an so that that's fine but really know what we wanna do is actually once was build those we actually want to use that they it it he added the exposing and we want to serve services and use those services OK so that's really what interests interest the I so under that talk about these really knowing details but what I have is that I have some friends in our in Switzerland and his colleagues will scale and I play a lot with it because it's in production the it uses call stack up behind it and it is a it's a fairly basic cloud in the sense that just for you just instantiate and create virtual machines and it gives you a virtual machines are kind of like Amazon which means that it's a public AP and then you manage the group and keep the soul of things I'm actually going to show you some some demos and and we can use that about because it's in correction and my point being that even even though we are building are we're trying to figure out how to build those clouds for within companies private province on there are some of those that are not easy to the power of the interaction of up so class that originally it was just got static and I just modified it helps stacks of because it wouldn't be fair to our knowledge is due to the prospect but you'll see what I mean by this is that if we look at all the components of those uh yes I of course they have you know something to handle compute Sovietization farmer hypervisors and they have something to handle storage whether it's Image catalog or a data volume and and networking with heard about it the and and and things like that so those are the 3 main no components there then you have the image management so you know in in class that we actually don't give a name is just the image management are in OpenStack that would be against our identity management standing class that we don't give it a name but in OpenStack there would be keystone so compute would be Novak controller but for us so that we don't have a name but you see that you know in all those project where you take OpenNebula OpenStack clouds that you cannot find the same thing and those 3 exposes an API and on top of their that they have a net net CVPR and will talk more about it or they also have an to Amazon bridge and that's really what we can used to build those additional service batch more so terms the best word you know for example is just example to show you that all those components disappear all systems have the same type of of components and that's the Dutch word for CloudStack and Chinese because I worked a lot on the look at position i to get OpenStack you know there's another dashboard called arise and and of course if you look at open that they also have a Dutch that's called still sandstone and and it comes with so you gonna find the the the same things in all those yes solutions in terms of API that's really that's really the key and to me that's that's where the cloud is about it's much more than 2 transitions really exposing all the campus capabilities of those so system through a easy-to-use API so class that has a very extensive at the Ivory functionality exposed to that API it's not a cure REST API it's such stupid Jason but there's more than this to rest and just TPJ isn't because it's more of a query at q of OpenStack is sparser normally fully REST API in the sense that you you have very well-defined URI and then you can use different types of methods GET POST PUT delete to actually manage those resources in class that everything is going to be a ghetto post that so they have they have a kids so that's fine so you build their cloud makes for the nuclei you happy you can start playing with this OK so now how do you do this well the button you have and a and you look for clients so you need a client to talk to that apply to the crowd I left side is OpenStack you do OpenStack clients to get 72 repositories and give higher and then class that you you call stack land you get 22 and then whether you like and I would be guy screw you pick the 1 you will get the only follows then you see guys coming in and passing for so you know Python lots of Python in in OpenStack of course it's all written in impact on also have Python clients in In call stack so really what's interesting there is that whatever language you're using whatever at at you trying to do you going to find a client for that cloud on on data state about clients you know we can debate whether there is something happening there but so let's look at a few clients of the next 1 is specific to crops that but then and I talk about a few 1 within the aperture suffer from nations that actually are common to all to all of the yes solution but also all the cloud providers that don't provide self remember that provide the counselors but Cloud monkeys specific prior to class that that's really RCI so C 2 has a to see I of course that unions value machine and that's how you manage to true I have supplied monkey that for us that's 4 parts that you run parliament can and then you have access to all the API is of parts and that's how you you do very low-level operations in in your problem now of shoes on the so as easy as it is that monkey if you wanna human but now of course I work on debouches suffer from so I like all the projects Landsea DSF there's tons of them you know 150 projects the SF at cloud is very interesting and I work on the ground because it's in Python and I do pipe pounds so even though working class that which is just the you know I could be more in the cell I would use usually the applies it's actually records a Python wrapper on top of all the API not only class that OpenStack but also you see SoftLayer the GoGreen Dreamhost Amazon Google Compute Engine and so on that means that if you're using multiple Clouds while you take the cloud and then you can write an application and basically the differences in API are going to be totally abstracted and you're going to be you use that that would you some idea of the starting to see that we get those CRA exposing API is now we have wrappers Our and of course some of your thinking standards but we'll talk about that but is so typical the props script I looks like this this is by upon you import the modules and then
you get a driver for that class so you see exo scale I just instantiate the pixels cal
driver specify my keys and then I have access to a list of commands that are common that are the basic PIV of plot so this image this site is those are very basic functionality of every cloud no is the is the zones is the templates this the instance types started no things like this at the bottom you see here is a deploying node method where you can actually start an instance and then passed strips and execute those scripts on the on the deployment of the typical the cloud onset the and this is even light exo scale which is my Michael Stipe if you remove that the driver line OK all the rest is going to be valid for any cloud OpenStack OpenNebula you see 2 there is nothing specific about constant well almost the object of another 1 that who uses out yeah OK 1 person right nights heavily used J. crowd is heavily used at another project at the except it's a popular project the they they graduated since the level project uh it was donated recently to the SF sol-gel and basically it's just like the Cloudlets and Jarvis interview you Java library that can talk to any of those clouds and abstract differences in the nucleus and it's used in lots of tools that you may use using without knowing the really that there's Jacob's for example if you guys ever heard of cloud fight From GigaSpaces choosing a Jacob's Delta cut is another 1 that would be that's in Ruby and it's also SF so here you see that in the the club doesn't matter because you have wrappers and then the language doesn't even matter because we are fine the client and you just at the SFT of 3 projects and do the same thing and that are in the language of your choice so that the cut is in Ruby are the bit unfortunate it was kind of sponsored by right at the start of buried at and then they dropped it it it it seems odd that it's the same thing it's gives you a wrapper on top of all the the differences so now that's great but you have all those interfaces and nothing is standard so there is no standardization on the API and then you know there is no Alderson wrappers are are different so we're seeing some standardization effort here uh from the and yet we have this see standard that's coming up which is an attempt at standardizing all those API for the cloud and the CCI is another 1 I from Open Grid Forum or right away you see the problem that we're talking standard and I'm leading you to so that's Self problematic and to marry the standard is Amazon that's the de-facto standard and now you know it's not that they're not pushing it did not pushing for it to be standardized but since units so heavy use that's the the de-facto standard is gonna be interesting to see what's going to happen with the Google I Compute engine that just 2 was a was made GA up which is yet again a different API and we'll see that that gained some steam and and and becomes more of a standard now OpenNebula for example is pushing really hard to have PC to compatibility of course you can read this was from the start and she to clone so everything to do is this a 2 compliant with Amazon actually if you would you tell then you look for a eucalyptus client you'll find 1 is you go up the but in our in class that what we're doing anyway to recognize that the C 2 is very important so we have an easy to bridge and we are developing a new 1 that to get there and then we also have a Google and compute engine bridge which means that you can use the Google Compute Engine clients to talk to concept but then so we get that cloud whatever it solution you're using exclusion API and then you choose your upper OK that's fine but now we need to do things on top of this up so of course you 1 of the my managers said you're going up stack I don't really like it but anyway that's what's happening on top public clients you know you're building up those services that and that's really what we're trying to so we going towards the Platform-as-a-Service and I and I took a clue again from the Amazon Web Services I because I knew about that cloud formation service and then I also yeah knew about obs works but I don't really understand the difference so if you go to their page that application management page the nice if A Q and is this picture where basically it starts from the real level services easy to start an instance of a creative firewall critic keypair things like this you most sends very basic image-management lots of control and then you go towards building templates orchestrating orchestrating multiple instances now so you defining the software running in those machines you're defining the number of machines running how they interact and then you go even further until you reach being start which is really now for your app developer who is going to do think that have been just stick it through the PI and as the to the to the magic so somehow we have to find a way to go from yes API clients and then move all the ways that we can we can offer those very nice bus services so we have tools that going that direction I something like Apache were here the data right now but it's so it's Internet that she of course is based on Jay clouds and what happens there is that you write the definition of the the the and said you want started the software you want on them you can use basic bash script for configuration or it can use the conservation management tool the and then you start that that cluster of instances and it's going to be provision that me that let's say you want to build among would the new cluster on ElasticSearch cluster to define this in in your file you object crowd template fitted to to to we're and is going to is going to start so that's great but but what we're seeing also now is from the the guys in our configuration management but I think we're seeing very very interesting things in the conservation management room was full of less that so that's just an example we shaft but again you could use you can do similar things but that and stacking and sensible forward on so here with their and developing on my laptop by a chef-client and I define my recipes for the software that I 1 installed in those 2 machines and now it's going to happen is that from my laptop I can actually use the the nice tool from from obscure to actually did directly deployed my node in the cloud and it's going to install the software that I that I find so I show you so you um an example of this that thing that's that's very very powerful because now you're bridging bridging the gap between configuration management and actually using the crowd to you on your laptop you have all those recipes you push and suddenly you have you know 10 instances with whatever you want know Elastic Search how whatever and the problem was becomes a transparent yeah the next 1 this is great but lately I've been having paying vagrant a vagrant up that comes and the grant is a tool to work basically created machines and initially to credit emission a locked up so that you can now use configuration management whatever you decide no causal back again show but that simple bash scripts and you can we can define the software you want those machines in a on fire and there's going to create the instance now what's interesting that vagrant is coming up with the plug-ins for all those clouds so now you are developing and at the same time directly from vagrant you can push those instances in the cloud and build those infrastructure because take a break incidental I feel like rules in new so is more at that what you wanna see OK let's and
should I know it so as a cloud monkey for example so 1st let me show you angle scale so I have my cloud that seems a scale it's very this kind of network it's got stuck behind it doesn't look like us that because they really the entire UI at but as very basic functionality were great so you can start instances defined security groups pairs things like this the there's no guiding the demo
if they can be on the net a new so here I have 1
instance running as I said you can define secretive groups with insecure groups you can define in grass grass rules good so cloud monkey here is connected to its configured to use our use this exile scale Cloud so now I can type for example of the type list that bad and it listed all the list type API pulse effect so all those are itchy eyes are available that I can use to government and so is volume is the PC the the what can we do yeah yeah of course this virtual machines OK it looks like a bit like
garbage that you can shoot the you can filter that our I mean so you get myself the other very nice like your father you can you can start messing with things with the
cloud so the if you know by from here and just entering her and I on interactive shell this and basically I have the I've initialized it with some connection objects so i have a connection which is the nanoscale scale driver and it has a bunch of methods create there create node you name it so if you do for example con uh these nodes so you see the same the the same node you got can't images it can you
see all the images available on its EiconScript Andrew interesting
in interesting things out let's now it's try to announce Hadoop cluster
yeah so I mentioned this tool from SF were Apache words or accuses GEC crowds to talk to the crowd and we go I give it the configuration file but so is just lounge cluster and you give a configuration file that just as list stuff secure to groups you want to keep their you want the software you want installed version so here I'm saying I want that Cloudera Hadoop distribution and things like this OK so it looks like it's starting well so it's start 10 nodes I and then have 1 name had to name node and then jumped records so basically it's using the clients starting the instances and then it's going to SSH into it run the bash script that 1 of the basic version of running batch to install had to and had an advantage to return and the mean of the SSH sequence to get on there so you say look at that the same back at my best for here looks like it's starting on so I got all my had to know that are starting when 12 4 once they're running all green and as essentially to connect and stop consonants yeah the 1
that I like a lot the chef so CS service when I say working with virtualization the you know all the the all the users said all I want is you and I 1 of the and Y because I want be route
OK so here's iteration you root and then after couple months they're like 0 no I don't 1 you with on the software give back scenario people are back again of using configuration management and not going to that's interesting so anyway so nice you know you can do the same thing was nice about using template this so all those tools give you the same capability whether it's the class object Kraut
knife you know you can you can do things all the templates are available but here will gonna do just for the heck of it uh is this so that some black magic can you can you read that yeah the so nice server creates I wanna create an instance of a small instance died time Warner use will win through as a template I went and when that starts and into SSH as root using this scale when you start being stance take this key pair which is preconfigured used this security group Elastic Search I know public AP that's that just saying they just telling knife to not locate an IP address and try to consider and act and things like this and then here that the configuration magic where I I tell chef to up to try to install the elastic search had a role and then I do the so the name for them let's try to stop that but so bad role what's what's in that always is basically a list of recipes that I want shift to to go through to configure the software and I pre-created this that's actually the Elastic Search recipes coming directly from the did have a ripple so hopefully if things go well that's does the instance that's going to SSH to that instance installed a chef-client then go through the recipe and then if everything goes well I will have a working Search but with the Nanjing Stenger next proxy running in 5 minutes the FIL we met the case of doing that time I might add clusters bootstrapping all things are running I see my force them machine has started in a different secure to go phonetic search how is that the always
chickens running the subject of his bootstrapping setting and lots of rules for the on the trials a so we let that they will let that go and now and then I'll get back to no so I was talking about you know
vagrant and grants for their workers but
I can see approaching guys are actually starting to
use their grants because you have configuration management in there that you can share and then you have plug-ins to deploy in the in the cloud so that's so they file and you see again you specify your of your box you specify the post the keys and then what's going to happen is that going to push that machine so those are kind of still low level you know low-level tools that you can know that you can use the differently when we have is that all the out there club from the addition of the the the beta uh in in as a scalar slipstream which are less known R and and codify which is doing quite a bit of traction and all those work with cross that Paul OpenStack I believe their workers were yet OpenNebula probably so those are some pairs now this you know going back to the C 2 and we have that you see to service we're trying to rule it out to the left I haven't clot formation can you do clot formation with any of those Infrastructure-as-a-Service solution think of that we now have fact tuition at some give have it's a Rails application of so you can actually use the back end if you move your level and you want just push all CloudFormation template you can use the back and part of this which is called statement but otherwise the expectations was the view of the web from and where you can enter your transformation templates and then the the stock but of course OpenStack as he which as I and I couldn't find a good picture of heat so that's the 1 I found on Google as I understand it uh OpenStack he can also consume uh CloudFormation templates and is also does that will to scale OK so we're starting to see those higher level services in many of those solutions yeah but so let's go back to and see if we made any headway yea so this is
where so if everything goes well yeah and way to SSH this guy which is the Hadoop NameNode
OK and now I should as Hadoop running the hopefully yes OK and I run a MapReduce jobs but I want you to trust me that it works OK I so you can go and actually put data uh the effects commands and then actually run that produce so right now I minus that's something with the security group to see section about work but it does it does work so I need you to trust me on that so that's fine so we like where was that 10 minutes i provisions in a Hadoop cluster in a crowd I think that's that's pRecall why am I trying to do this because I want expose map MapReduce API that I would like us to be able to provide an Elastic MapReduce service on top of any class that care cloud so that's the that's the bases that you need to start with you need to be able to start a Hadoop cluster on demand and then you know expose your on our whatever API so what's the other 1 so chef is still going 1 this shifts to go on yes shifts to so to wait for
chefs what we gonna try to do is when try to start another node in in in the account there's the harm is a this is boring topic if so that would be con collection you know deploying node a deploying node you give a name to your instance so as them 1 specify an image the specified size OK and then use the society name has that keep their so is a scale and then specify the SSH key
where is the private key for that keep their that you're going to use the and then you specify a deeper scripts the poor people and as the so that's the type
of command and that's the that's much lower level than much lower level than word or although the the 9th common
OK so my 1st finished so if we go through everything and you know bootstrap in Seoul chef and then went through the recipes and then studied and genetics because if i expose the ElasticSearch cluster are that's that's not good so where is that was that node so it's a for them yeah so for them IP that's DAP then you can do puro for elastic search or chromosome nice uh
extension which is Of course since so just this side the I P of that node 80 80 because that's my next for on this set of the a set of basic a
publication but I entered the login and password before so if and went it's writing so do what almost perfect general but it's good that is gate was has to
search engine X running still doesn't like something OK
that was working How but our node here
OK cool so that the players finished here now what's interesting with the with the cloud is that the when you deploy with this that's just basically a list of scripts and now I should have so I the standard out
of when I define this it's a list of script 1 of 1 of the script is called scripts and then when finish is you can get Senate out from it and when I was running was being date OK you still have a world of grid computing yeah so that works we deployed a node then and ran some scripts on OK so
now so let's finish with my oops the where yeah destroy cluster considering small there's just destroy everything at some demand so being and when we should see over here is you should see all the instances stuff if the in a the it that's not the part I was expressed to have problems but the the network is in the case of the network you know probably in a problems in it yeah OK but that's that's
disagree it's of at the get the idea that you know we're trying to build was higher services there is a choice of so clients and and tools on top of that and whether your Python Ruby Java guide you can find your your client for whatever solution cross tackle insect OpenNebula and then we're trying to expose in Ohio little services that that clone Amazon and we already have formation are so thinking we can expect more like Elastic MapReduce beanstalk and sure I get involved that he got that look it up and that's it thank you be thank
Prozess <Physik>
Punkt
Formale Sprache
Versionsverwaltung
Bridge <Kommunikationstechnik>
Kartesische Koordinaten
Computerunterstütztes Verfahren
Computeranimation
Übergang
Streaming <Kommunikationstechnik>
Client
Skript <Programm>
Auswahlaxiom
Addition
Lineares Funktional
Datennetz
REST <Informatik>
Güte der Anpassung
Gebäude <Mathematik>
Abfrage
Systemaufruf
Biprodukt
Kommutator <Quantentheorie>
Dienst <Informatik>
Bilddatenbank
Client
Projektive Ebene
Information
Cloud Computing
Schlüsselverwaltung
Aggregatzustand
Standardabweichung
Subtraktion
Gewicht <Mathematik>
Ortsoperator
Data-Warehouse-Konzept
Stab
Gruppenoperation
Klasse <Mathematik>
Zellularer Automat
Interaktives Fernsehen
Online-Katalog
Dienst <Informatik>
Analytische Menge
Term
Virtuelle Maschine
Datensatz
Software
Speicherbereichsnetzwerk
Datentyp
Wrapper <Programmierung>
Zusammenhängender Graph
Elastische Deformation
Spezifisches Volumen
Speicher <Informatik>
Softwareentwickler
Bildgebendes Verfahren
Modul
Leistung <Physik>
Open Source
Physikalisches System
Cloud Computing
Modul
Kern <Mathematik>
Packprogramm
Quick-Sort
Energiedichte
Mereologie
Identitätsverwaltung
Wort <Informatik>
Stapelverarbeitung
Klon <Mathematik>
Bit
Formale Sprache
Bridge <Kommunikationstechnik>
Kartesische Koordinaten
Computeranimation
Richtung
Internetworking
Übergang
Homepage
Client
Einheit <Mathematik>
Datenmanagement
Minimum
Kontrollstruktur
Skript <Programm>
Schnitt <Graphentheorie>
Gerade
Schnittstelle
App <Programm>
Lineares Funktional
Zentrische Streckung
Physikalischer Effekt
Abstraktionsebene
Template
Gebäude <Mathematik>
Disjunktion <Logik>
Plot <Graphische Darstellung>
Zeitzone
Konfigurationsverwaltung
Dienst <Informatik>
Rechter Winkel
Ein-Ausgabe
Client
Projektive Ebene
Cloud Computing
Schlüsselverwaltung
Standardabweichung
Instantiierung
Public-Key-Kryptosystem
Web Site
Subtraktion
Firewall
Klasse <Mathematik>
Zahlenbereich
Dienst <Informatik>
Technische Informatik
Service provider
Virtuelle Maschine
Knotenmenge
Webforum
Software
Reelle Zahl
Notebook-Computer
Wrapper <Programmierung>
Datentyp
Programmbibliothek
Skript <Programm>
Elastische Deformation
Softwareentwickler
Bildgebendes Verfahren
Pixel
Konvexe Hülle
Plug in
Mailing-Liste
Elektronische Publikation
Cloud Computing
Kern <Mathematik>
Objekt <Kategorie>
Druckertreiber
Offene Menge
Gamecontroller
Bus <Informatik>
Energieerhaltung
Zentrische Streckung
Lineares Funktional
Demo <Programm>
Gewicht <Mathematik>
Datennetz
Winkel
Computersicherheit
Gruppenkeim
Cloud Computing
Menge
Computeranimation
Portscanner
Gravitationsgesetz
Instantiierung
Soundverarbeitung
Router
Zentrische Streckung
Datentyp
Default
Gruppenkeim
Mailing-Liste
Schlussregel
Cloud Computing
Gerade
Computeranimation
Portscanner
Virtuelle Maschine
Magnettrommelspeicher
OISC
Puls <Technik>
Mehragentensystem
Ein-Ausgabe
Datentyp
GRASS <Programm>
Spezifisches Volumen
Transitionssystem
Instantiierung
Einfach zusammenhängender Raum
Zentrische Streckung
Nabel <Mathematik>
Datenmodell
Cloud Computing
Nabel <Mathematik>
Computeranimation
Objekt <Kategorie>
Knotenmenge
Druckertreiber
Einheit <Mathematik>
Parametersystem
Druckertreiber
Transitionssystem
Bildgebendes Verfahren
Distributionstheorie
Folge <Mathematik>
Gewichtete Summe
Versionsverwaltung
Gruppenkeim
Nummerung
Computeranimation
Client
Knotenmenge
Datensatz
Software
Total <Mathematik>
Statistische Analyse
Skript <Programm>
Strom <Mathematik>
Widerspruchsfreiheit
Gammafunktion
Mathematisierung
Mailing-Liste
Elektronische Publikation
Schlussregel
Arithmetisches Mittel
Konfigurationsverwaltung
Office-Paket
Emulation
Strukturgleichungsmodell
Einheit <Mathematik>
Wort <Informatik>
Stapelverarbeitung
Stab
Zentraleinheit
Instantiierung
Objekt <Kategorie>
Konfigurationsverwaltung
Dienst <Informatik>
Virtualisierung
Einheit <Mathematik>
Software
Template
Klasse <Mathematik>
Iteration
Routing
Transitionssystem
Computeranimation
Public-Key-Kryptosystem
Proxy Server
Server
Hecke-Operator
Gruppenkeim
Netzadresse
Computeranimation
Virtuelle Maschine
Magnettrommelspeicher
Software
Massestrom
Elastische Deformation
Wurzel <Mathematik>
Druckertreiber
Verschiebungsoperator
Zentrische Streckung
Dean-Zahl
Template
Computersicherheit
Kanal <Bildverarbeitung>
Dämon <Informatik>
Datenmodell
Netzwerkbetriebssystem
Mailing-Liste
Marketinginformationssystem
Zeiger <Informatik>
Gerade
Inverser Limes
Netzwerktopologie
Konfigurationsverwaltung
Orbit <Mathematik>
Forcing
Wurzel <Mathematik>
Einheit <Mathematik>
Dynamisches RAM
Server
Hill-Differentialgleichung
Warping
Transitionssystem
Instantiierung
Transinformation
Konvexe Hülle
Konfigurationsverwaltung
Information-Retrieval-System
Gewichtete Summe
Schlussregel
Gerade
Computeranimation
Gesetz <Physik>
Wurm <Informatik>
Schiefe
Menge
Ruhmasse
Firewall
Hill-Differentialgleichung
Bit
Quader
Kartesische Koordinaten
Transformation <Mathematik>
Nummerung
Computeranimation
Übergang
Virtuelle Maschine
Service provider
Erwartungswert
Front-End <Software>
Datennetz
Addition
Befehl <Informatik>
Sichtenkonzept
Template
Güte der Anpassung
Plug in
Cloud Computing
Konfigurationsverwaltung
Dienst <Informatik>
Wurzel <Mathematik>
Mereologie
Dateiformat
Zeitzone
Versionsverwaltung
Normalspannung
Schlüsselverwaltung
Soundverarbeitung
Synchronisierung
Konvexe Hülle
Computersicherheit
Klasse <Mathematik>
Gewichtete Summe
Gruppenkeim
Farbverwaltungssystem
Cloud Computing
Computeranimation
Mapping <Computergraphik>
Turtle <Informatik>
Physikalisches System
Dienst <Informatik>
Verzeichnisdienst
Prozess <Informatik>
Ruhmasse
Stützpunkt <Mathematik>
Garbentheorie
Bildschirmsymbol
Elastische Deformation
Verschiebungsoperator
Public-Key-Kryptosystem
Zentrische Streckung
Knotenmenge
Datentyp
Datenmodell
Skript <Programm>
Transitionssystem
Druckertreiber
Schlüsselverwaltung
Bildgebendes Verfahren
Computeranimation
Instantiierung
Server
Bootstrap-Aggregation
Gradient
Computeranimation
Übergang
Spezialrechner
Knotenmenge
OISC
Statistische Analyse
Protokoll <Datenverarbeitungssystem>
Strebe
Wort <Informatik>
MIDI <Musikelektronik>
Elastische Deformation
Druckertreiber
Transitionssystem
Knotenmenge
Verknüpfungsglied
Menge
Passwort
Maßerweiterung
Login
Computeranimation
Physikalisches System
Knotenmenge
Synchronisierung
Einheit <Mathematik>
Suchmaschine
Adressraum
Information-Retrieval-System
Flächeninhalt
Extrempunkt
Computeranimation
Datenmodell
Skript <Programm>
Mailing-Liste
Computerunterstütztes Verfahren
Transitionssystem
Druckertreiber
Cloud Computing
Computeranimation
Datennetz
Synchronisierung
Extrempunkt
Computeranimation
Graph
Einheit <Mathematik>
Objektorientierte Programmiersprache
Adressraum
Mereologie
Phasenumwandlung
Skript <Programm>
Cloud Computing
Druckertreiber
Transitionssystem
Instantiierung
Client
Dienst <Informatik>
Applet
Dateiformat
Elastische Deformation
Elektronischer Programmführer
Auswahlaxiom

Metadaten

Formale Metadaten

Titel You have a Cloud, now What?
Serientitel FOSDEM 2014
Autor Goasguen, Sebastien
Lizenz CC-Namensnennung 2.0 Belgien:
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 10.5446/32673
Herausgeber FOSDEM VZW
Erscheinungsjahr 2014
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract You have a Cloud, now what? In the last few years we have seen many presentations focused on how to build IaaS clouds. However very few, if any, actually tackle the issue of how to use a Cloud once you are done building it. In this presentation will look at key open source software that form the cloud ecosystem and are used to make use of a working cloud. Specifically we will review software like apache libcloud, jclouds, delatcloud, hadoop. We will also review the state of configuration management systems and their support for IaaS cloud software. We will go beyond talking about Cloud APIs and focus on API wrappers and how they are used to automate provisioning of virtual infrastructure within IaaS deployments. These software typically are not key to building a cloud but are used to make use of a working cloud. The higher level software are key to build applications, workflows or higher level abstractions like PaaS. We will go beyond talking about Cloud APIs and focus on API wrappers and how they are used to automate provisioning of virtual infrastructure within IaaS deployments. Specifically we will discuss API wrappers within the apache software foundation: libcloud, jclouds and delatcloud. To complement this wrappers we will also discuss OCCI and CIMI standard interfaces to IaaS systems (CloudStack, Opennebula and OpenStack). We will also discuss the state of configuration management systems (Chef, Puppet, Salt and Ansible) and their support/use for IaaS cloud software. These systems are of course used to build clouds but can also be used to provision virtual infrastructure within these clouds, the latter will be our focus. To illustrate our point we will describe several scenarios where we use Chef and SaltStack to provision a hadoop cluster, an elasticsearch cluster and a RiakCS cluster on a CloudStack based cloud. People attending the talk will see how the open source ecosystem build around IaaS solutions is used to deploy applications and workloads that they need. They will see how a IaaS gets them to the point of thinking about the applications and a faster provisioning time for their workloads instead of thinking of just building a IaaS

Ähnliche Filme

Loading...
Feedback