Bestand wählen

Designing NRT(NearRealTime) stream processing systems: Using python with Storm and Kafka

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Erkannte Entitäten
welcome back everyone once again I'm I'm also something from 16 or in Norway and I want present you cannot model we're talking about near real time stream processing in Python with storm in Kafka it was good morning all of all I understand I'm the only 1 standing between you and lunch so I try and recognition we all right so the title of the book is designing you're right the invention present systems now what what do we mean by data pipelines and nearly event systems so everybody of us knows that in the year they during the of information as and when we visit that site bells wrath applications and stuff like that and we need reliable and quick mechanisms to collect data once we are able to collect that data we need to process and analyze that the that might not systems like group that efficiently do but that's again but isn't system when we got what but systems we mean that this will be high latency of updates in your query on and then you have an obligation to the 1 of providerecommendations you wanna provide better features to users who want to provide a feedback loop lands and you want to layer which helps you work with the same data and update the dual latency of object so that is really a sense of the nearly invented by things work on how likely data is majorly into the parts 1 is the input that is basically the missing there you collecting all your use of user data that is coming on in streams the 2nd is the process in the region of possessing the reason the public not that I cities and it defines set and minutes for entities such as defined moment of development process on your community event stream flows in the normal process 1 month be on manually event processing right and then then a process that it out there has to be a little bit like imprisonment it up in probably agree with Wal-Mart or format that my other applications can consume enough feedback so today's talk is majorly focused on 2 components 1 is the missing there that's Kafka that and indeed in the 2nd part of the book
and before that goes over about about of open source project which was open source but without is now about to project known as a participant primarily about storm is 50 % close and 50 % but that doesn't stop us from using it in buy things and that is why I am here to the present how we can use strong and its capabilities for obtained using Python it's so how would try to structure the book is just before that can have a show of hands how many of us actually know what's from all and only of us know what software is right so we have a
majority of people that are still not known groups from encompasses just provably for what storm and look but given the time frame and the kind of thing that I have I might speed up speed up little bit but do you think that's enough to go and I'll be happy to see them was announced alright so then we got what stream processing them off challenges at that you need compact but this light on if it's a few of them 1st of all for me stop the system is actually an infrastructure that we
a never ending data-processing for me right so what I mean by that is it's not like a MapReduce job that at a given point of time at the finish line of the event processing system for me as it did keeps on coming in huge volume at 50 processing it and keep watching it had right at the level and for me so that is what I mean by the union the same data processing biplanes right you have restored Mrs. somewhere before your consumers and consumer you want you might want the player misses again you might want to drop to masses and stuff like that so all these things need to be taken care of by the use of nearly adventures you need a router messages for example you collecting you miss from application 1 you Mrs. from application pool you want your applications 1 Mrs. local 1 particular woke because Paul's Mrs. local particularly for different local so these few answers that people around the year they made up by think and system by your skin for high throughput for example you design a system that is taking care of thousands of MIT's history but what you might scale and you might want to use the system which was then pass about 100 thousand of message for 2nd right so you need a design philosophy around a system that takes care of all that then we talk about stuff on it it takes care of almost all these things for example you stop and then you have already sort of variety of Saturn is that you can see for example you just want to
process all the Prokofiev that you're getting from titles was and ambiguity all right now you and systems like this you need you need to take care of the back pressure for example of what is my for assessing their dies of it 1 my misses school right that coming from the models on the final map persistently it is done so how store personally how does the processing they're taken off the ahead and what happens to those messages right so produced on a lattice from the 200 extremely but brought this scenarios for example you would look beyond flight and again that has to be distributed across the quantity that distributed across politics are right so about somber something known as the RBC before that had that that it is scalable so when we go in the architects of storm you'll understand how all of what we mean by scalability of about a strong so you have different services that difference of Boston strong and that it's scale is that they're going to talk about scalability we obviously don't work for tolerance as well what happens to 1 of the workers if it goes down more happens to 1 of the sources that coordinates all the world is modeled as I mentioned earlier I at any given point of time and what what might be the president by 1 of the problems might go on they might come up but In the end I don't want the whole system aggression for me because that is is certainly not feasible it is programmed that language-agnostic that is how we are able to use Python over the part it supports all other languages that so as I mentioned before is inclusive and but it has the that help you achieve your lot stuff there are fewer once so level off from the kinds that we have slightly reduced by Python but I believe also in the word map that man will open up open that up using the modularity is that achieving parallelism up across components so when we understand the various components of 1st on and the about how at each level we can configure what the barrel is moving so these are the song components I've tried and depicted in the pool parts 1 is the conceptual view of this conceptually had to when you actually programming so just system and then is the physical view is that the actual sources that will be running on on on analysis so was on a on a system trained so for us it's spouse OK beats followed as ingestion unless it's ingestion component so spouse isn't interface the of the world right so you don't like it follows that basically interface
with the with the firewalls order of eventually or carefully on anything like that and then spout basically generate stream which we will talk about stream is nothing but the unbalanced equal of couples that is to be good bolts and other other components so here in the example why financial orders a small good effect of smoke now it's doing nothing in in the real world you will ever come across a because what is doing it it's not connecting the alcohol and it's just generating random sentences within itself and and making them ahead for the for the processing right so what exactly doing is it says I want generate lands inventors from the list of sentences
that I have and this and that each sentence by schemes to talk about so if you can see In the court's not what's yours is I'm meeting each sentence right so stream is non bounded sequence of couples so each sentence is part of the screen then so each message in the speech is the emic method that's coming from the spot now these streams that come board so OK so there's a fine-grained lines that you cannot see so basically board you can have many a number of inputs and number of outputs of inputs the extreme as input global so board can take input from a spoiled aboard can take input from other spokesman aboard what's output can be improved when the Board of all output can be a persistent store on anything else all about my produced from anything that might be they enjoyable quality of how we need all this together is what we call a topological circle of you have my deepest quality that might be available what's going laughter so as mentioned style is just Interface for Outdoor word to get it and and will respond by boards are essentially processing this very related projects he was like food things you have a good idea it's not it's not that stuff support essentially that the whole architecture magnitude together we call it a logic so we'll storm plus W always some medical not be flexibility is such that you're probability can be a mixture of awe technologies related to you might right this that written Mandela you might buy that Python and that's that's really helps you a lot because the example in my vitamin we a mixture of Java and Python that state so I would all this is not create something that is already at an uncertain and all of those guys might not be labeled trees has been written in python sort definable broadly has how we define the scope know that might be released without because there multiple libraries right and that helps you predict which strong this is an example of stream past library that focused role the property so this is actually a
closer use so that sort of thing that has been defined my Python spoke and my beliefs so what what is happening when you just have to find that I have awards follow so this is actually a board which is taking input from what's out we get into the meaning of what's happening and I thought this board that the program that need to and the branding
even input is the sentence that so what it's doing is it's receiving sentences and it's spreading the sentences in the woods the next boss receive the
input from sentence because which essentially it's getting the words from the and over here ask the it's revamping groupings what things mean and then the vote count there is essentially nothing it's taking inputs from the previous work which is the work on board and then processing it as a mention it's a dead end it's the last word in microbiology it need not anything you can make it mn anything on the need of making a meeting in right so this is what's happening in a basic quality so what we mean the groupings now so groupings will help you back and translate you group by a particular user and then processes data across 1 look at itself because it's a distributed system if it is negative stupid you might have probably ideas
right so a case of simple work on example now you just follow the the meaning of words and borders processing and at the end of it you want councils conf by words right now 1 we that is you have a persistent letters dumping everything over there and that's basically said that it science yeah mechanisms so you're watching what that and I and symptoms of ridiculous that in and of the board what and was and group them by what's on the board itself so it makes sure that each word seem what was on the same task and they're demanding ideas for the the other shuffle suffered make sure that each board received the same number for the same number of streams but they might suffer across a number of them are grouped by any any from into the and there's a lot of other grouping events and also the documentation has a complete list with the good examples of coming back to the physical you in physical abuse but 1 I'm sorry about the size of the data and so on the extreme left you see numbers some members if you coming from a new word from members just like a job practice what you do with
nemesis you submit jobs to the most members make sure that these typologies are then from across the a so members we think of managing and monitoring your booklet on the cluster you submit a new topological it takes care of the deployment their task is assigned on gives a figure of a task the takes extent of the assignment no seconds later had mentioned for the confidence value of the 1st one if my muscles down Michael below the state depending what what they get infected as the new goal others will be what arguments some the new book religious but the top is already done in the logic effected by that so so it's even if an imbecile is down because it would want to answer but because was intact zookeeper is the coordinator between supervisors and a bus either on the actual so this is landing on the local that takes care of the probabilities not and but communicate by other people's to give it is nothing it's a it's a kind of cluster configuration tool that helps you manage what cluster stats are so that members and the can be again if possible ways so somebody has few parts like workers this has tasks and you just so metabolizable is known it could make sure that another will correspond so that in that this for tolerance is maintained at each and every left come into the many How
Stallman by converting together 1st is the past that's a library and will probably the library Open Source by fossil energies in the core open source of view about what the Boston Deputy stream past so that lets you write your book broadly let's relates faults and Bolton Python but you have definable broadly final definition of broadly using the through the use of the advantages of that is you can then makes multiple language for example jostled and Python balls of that is why I prefer writing that included yourself but if you want write anything in your Python there's a library of spectral that was open bundles various last year so that it should complete apologies spouts schools everything invited now we have another government affairs which I believe is more about what inducing using that incident as so is the world talking about that so he has a library what class that is basically takes care of slow mixing your job and your and stuff that's even interesting library so I think you should go with that office and the Native Java project of storm itself let's unite people produced so but it will not let you define a global audience by then you have to write code your put your overloading dialog you can both invited and there are few implementation if people have written faults in Python but that are not that successful so that's the last option I would say you know you can use violence being but so this is this is the architecture tonight stream pass if you so all the great part is what is cluster right now what's the what's that mean is is it lets me create an environment has almost would storm cluster alright so I am on my bed machine and I'm creating this probability found fighting bullets and defining floats on water needed was so far as fast as utility that comes with low the impossible and so when I say sparse quick-start a particular project it gives me is beautiful privately owned by each forward on each side has significant so far as fast forward with it lets me come configured on my will close by any book of this basically means something to stomp apology while fast last the bus it basically does the it's it's to each of the local and the deploys your Python environment all of the so for example if I'm using something like gold that topic line or of the client that means it's external dependency that other company to fight them and this will make sure that it 1st of all little local and the 2 environment to install on my dependencies Howard Fineman dependencies and modifies then and you could say that these are my dependencies so it will be in what'll environments and you created quite stir each defined over that global of these is better refined through the BSS subproperties last Evanini wonky political broadly you define your book review this is an example of a book so it's it's a lot of about 2 years if I sorted out instead that I have followed which is what struck me when I have a board which is what conduct the but it's possible if that's what you see in the bottom of the bottom row is the pool which is essentially defines the barriers and that it would be to perform therefore flocking I've been 1 application of this so that I would like to focus so the application is built around the key mediated planes what is happening in the key given it was somebody it's an article that is the integral for that so it's really think of what for about is the law of of
cooling political capture that feed from wikipedia cite this taken 1 of the tools it gives me a fine distances this looks like this so basically you will think the mixes of the fate of forces will be action so many you believe inevitable actions entered the believed city at Harvard except but it gives the whether the article was aided by a book or not so is what follows is what rule of Italy was organ or not by making the teens of the users logged and you have a user name is not you you you you get an idea what you of so preventing it as the linear 1 is 1 the action life span so many believe how many edits how many updates there is 1 of the human listeners what the a graph of our modern world of human listeners of what an acid learning log users was as anonymous users now and what what's in the year there was the the level of sort of the interconnectivity so I think a subset of rocks from well from the beginning if you have done little to stop the both of them and I'm going start that after this time looked after model or so I did was I is that sparse quick GUI also underscored plants it gave me that reportedly that I was talking about after that this is about me and I was talking not all the 5 of them now what I do is I define this very simple coffers for I mean there's nothing tough simple parts of the news and current going and what it's doing is it's creating a particular topic and getting all the log files and emitting that message right so when the same as doing it's essentially isn't all that getting no this board basically passes medicine OK so odd others it checks what the action that is it takes the waters human architects where the users logged and stream processing and in form but so I would sit amet underscore making a list of words is and is a list of lists so basically 1 out of this would be a human G. Columbus with it got a lot going on around in OK now I'm painting he's a G 2 and G the head of government that would be delisted so that had the level of his final right so he's very very FIL this is our collective knowledge of what I've battling it is out of its orbital running than acknowledge that I've done the whole dictionary on the something and that lets me brought my body the also on the level of it so that the simple final word comparable work that I have so for this particular example this is the idea that all of and I'm saying you have to use best then the output of the copper so the path distance and once surpasses as understaffed baskets out become or sure this is what it looks like after pops into this linear
time parts and that is going on so this is the final word on that topic so that is coming in and these are the graph that 1st the
1st 1 is this created and I think the last 1 is deleted
then you have what was humans and finally you have log was in 1 of those so the point this demo is that it's pretty easy to get started with strong but it's very easy to diagnose faults and walls and finally you have hidden application that is coming and it does not take much of your past who managed to clusters that 1 of the 1 of the headaches that come with system that can
always you have to more also 1st Amendment with that but the storm system based on what really happened will lossless semantic and was the norm that wanted topologies learning it will ever felt how it really stop I have to manually collect apology the in the dining so as as in the laws that govern the way you get get past and that is what the meaning of the year made up by basically for me is so this is how you and it's something was passed around it starts opening senseless from cluster festival broadly I didn't
system once yesterday sure that this was pretty fine for you you can
was possibly and interoperability what group who was not all that global get submitted only to the members of the mistakes of shipping yard of course would you look at but in case of a sparse forced into sits on each of the local malls because we need to create a water environment for Python that basically takes care of the dependency so that's an extra step that is fostered otherwise it's similar somebody job at apology or Python the logic because it gets shifted was time laughter why we need systems they've got I mean we've got a lot of you already got so many missing their own we've got 2 of the universe is of or growth caching solution down so readers stuff got off the african and water stuff so for me the high-throughput mystery persistent missing system that takes care of parts of his that that takes care of high-throughput that takes care of distributed storing up money that is not in the early challenges that of mentioning the 4th 1 enables I needed was Mrs. left that couldn't that takes care of the messages and that in alright and also once have consumed as to what we reading in those messages because there might be other consumers that will be doing so when in order distributed processing that there might be a number of consumers that want to consume the same message right so consider the topic for example the QBD edit page topic no 1 can do much is eating those topics and there might be under consumer that is dumping all those to model right so what you need is you would like to build the mountains of the same messages you wanna autonomous system beyond 1 books for sale the low and high availability of data you want your data to be partitioned across multiple clusters so comfortable that you all that stuff along with that ejection reading hyperbole so you can have higher number of flights at high number of reads how would need to do that is because staff additionally is basically a 5 5 system-oriented cues like persist everything on this alright and by design and persist everything on this could start out on feature that has enabled disabled by the 214 you switch upon the system is so on the on the hottest so basically of installed everything on the hottest and let to repeat that by partitions and the biggest group was that you have these are the 2 important concepts for profit you getting into after this is all the the basic things they need to take care of clusters of color clusters basically cerebral because now you can have 1 rule single-order together property that single and multiple group can have multiple multiple encoding depends how you want in the area of a but after that you do that you even if you have to to say you are making it as an environment that lets you play your data across data centers is that something which is known as far from topics of public everybody coming from on that Mrs. about almost topic is just a linear this lecture group you Mrs. together not divide the stuff that is applicable consumers reading the same topic now each consumer can start from a different point and people are different but it's not like I say I have read and this is from 1 thing you must so each consumer in that it be prominent was 1 of the points that is what is known as offsets offset so in got so the wrong with you people you can maintain your offsets that in this topic has been led up to this point so it's simply like that the set that 1 partition is basically user 4 the internet you can partition of topics across multiple clusters so it's basically how it how it the thinking of how the works so basically in your environment those treated as log speak another lots and the environment that and logs will be used by multiple sources for example for some for the apple your pumping logs now 1 application is processing what you by analysis of application for the Singapore probably say RTD analysis you position that rule so it's the same what's been getting used at a lot of 4 different at Robert from how we can use start with Python they currently pool to implementations to that 1 is covered by them that's the that's the 1st model in the community and the other 1 is some so that's again open source told by the parsley I'm not sure what would in the sun so long because the Python works on the latest 0 was and obligation of some sort and coughed up storm really makes of year-to-year data by paying a very robust anaphora that isn't for you so if you're looking forward was implemented that the sort of Europe what other systems work independently so you can't you want not it's not a man to do stuff but strong cost on the cost of both of them 0 look pretty much independently so you are free to use any 1 of them that moreover once features of storm as well for example the the LPC still produced by natural log file and produced now then the new writing in the end of the year you can read a processing by by you have to make sure that each of the Mrs. messes get otherwise at the end of the your accounts we start with a but when we say each of the message gets processed once for and the vulgarity that it's not exactly once right so there will be limitations to this I will make sure that promises will get processed at least once and exactly once right so strongly to handle both of these things in the basic implementation of stumbled abroad that we saw it can only achieve you let you to the 1st part exactly once all right because it comes with a library EPA library the EPA does is it lets you acknowledge that the have process that upper or not if you're not process the couple at in that you do not go broadly into the play that particular top alright what we mean by exactly once is something that you need to figure out which electrical properties variable produced that I will only want to process the stuff it exactly once they so that is that is that the larger the brothers come the LPC is supported by modularity Pierre but in the current implementation of the multi like the triangle brothers are not supported so probably due to the have and remember I most of the from underlying do you have also that I use the following and is a pretty good so the first one is the official allergic to that of what is the bus library currently they're working extensively on that and they have to load uses ahead is well but in the current system or 2 a collection manageable Storm cluster pretty well the sample value was Cypriot logs from the usual conceived variables of foreign you get that from that you can submit a couple of years as we just saw all of have strong left as well they have softened by the libraries of Fisher documentation that pretty good football this system so it's a it's a must-read for of you and if you will understand how strong partisan works there's of brought by like it's it's 1 of the best blogs that helps to understand how and what's so what's in store just like a problem you can kind of discussion for ever a lot of progress probably and open open it up for questions just
before I called great I For what goes through the storm you where that comes with that is based on the way that they had to keep track of how the
clusters behavior so by some
miracle of the of so Michael Brodie was that began his work and that's been running since about 1 minute for the 1st time so the the the optimal book below the hollywood of those are the parts of the hearing on because between up from 15 minutes if you click on that you would see the different styles and different boards that we have in this so you have about and you have holes so I had once followed and that walls come constructed gives me how many resolving images and stuff like there's a difference between the government of the light is a difference between immigrant transfer because 1 of my bosses argument misses it's it's as simple as that topic was but thank you of this you and
what but yes it's still the case yes so so the question is earlier was in a strong blow up the topological we had we study but religion it's if the question so instead yes you have probably heard that my people the design challenges which does not let you do that for example now depends what kind of changes going resemblances going simple patterns and constant and stuff like that now it depends how the task within the get issue basically after being brilliant all of that is there is a feature request in which is known as long I think so so that its response so that we have to take care of this part it and you define your metabolism on-the-fly and you just with storm lost almost all that basically the balance on the order of cluster number of the new policies and that is fine but it was really interesting meeting on the mailing list says that the height of the side where . it very difficult to create systems that you can change them on the fly and then uh let them be the same was your columns you aggregates however misses have been grouped across different will close to really get affected by so I went and that's that's coming in the near future releases for is it life all have all the so the question is library EPA supposing that the question is dividend of abroad Python and but of the company and start filling sort happens to database the other 2 parts to it 1st of all of our they can probably go back reports that so this is how you define acknowledgment 3 years in a system so for that Pavilion might not might be because of a competent feelings
failure might be because of a business logic clearly in setting of the case I can explicitly field my and that you get the but yes it's 1 of my local word on the supervisory make sure study workers if was apples and using a module that is not available on what 1 of my vocalist stream past not thickness of medical graduates so this is how that we take care of them in case of a feeling of each in message individually this library EPA expected it to that with knowledge Porterfield either of the messages both on gives of business logic or any other scenarios that you have of so the
question is how we can steal the machines by adding more number of machines and increase the required so supervisor is walk around the locus of sorting to do is add more machines and sovereign supervisor where it will communicate with the Nimbus and then head how he was scalable yes you have that that's the 2nd 1 is coming you you have to restart but you and it will not take of the dynamic loads for that will you know that your data would be getting these automatic level right scale strong leftist on the common over that facility but it's very easy and unsustainable on that was start and you look at what already and all the task with the nutrient leadership had yes so essentially if is minus on on a system and that for GB and you would see different Debian's running on it so essentially what is happening is the supervisor of unsustainable local each topological method of locusts which in bed and have exhibitors and traits so each 1 of them and have a different process that is how you guarantee for tolerance of solid across multiple systems so what are the effects of spin again still the other components still for you so that is how this happens inside of insight about what you have been so OK so the question is how is on very different frameworks that we live in the market to sounds as 1 of the products that is open source by linking itself so linking nearly then those models for something that is 1 of the current open source know which is for unintelligible the spirits had never use it but yes there has to be some design philosophies that is where that is the existence of a few different systems that are there is 1 that is used by physical cannot that's open source of mark but I've never really given a short who either 1 of them because for me the lot of the components that we just on list the feature that this will really strong and they might be led EPA for me is pretty strong as well so did limit preferences for even more at from now I can't even the so close the question is how how does the exactly once and at least once this is what's so there's something known as of guaranteed possessing where you will flow flawed Mrs. across Europe abroad once that is done you work at least processing it once right now exactly once is what I was talking more than the advantages of probability that you that you need to write natural qualities which other abstraction abstraction over 0 the obvious physical qualities that lets you achieve exactly once so what at least once all your couples will flow through the globally really that is what's on takes care of negation events so yes there are a few notable storm in which it can make it so as you and you other other mechanisms I have to operate items without I'll be happy to see that after the that basically explain how he misses flows through which God different workers visible 1 board 1 of the board from spoken Ward and what happens when we expect the failure and stuff like that so the have to that the for
Schreiben <Datenverarbeitung>
Varietät <Mathematik>
Generator <Informatik>
Einheit <Mathematik>
Ordnung <Mathematik>
Lesen <Datenverarbeitung>
Selbst organisierendes System
Mathematische Logik
Virtuelle Maschine
Weg <Topologie>
Theoretische Physik
Hyperbolische Gruppe
Spezifisches Volumen
Cluster <Rechnernetz>
Offene Menge
Wort <Informatik>
Streaming <Kommunikationstechnik>
Prozess <Physik>
Mengentheoretische Topologie
Atomarität <Informatik>
Familie <Mathematik>
Metropolitan area network
Komponente <Software>
Parallele Schnittstelle
Konfiguration <Informatik>
Arithmetisches Mittel
Framework <Informatik>
Projektive Ebene
Varietät <Mathematik>
Web Site
Kontextbezogenes System
Physikalisches System
Inverser Limes
Speicher <Informatik>
Bildgebendes Verfahren
Elektronischer Datenaustausch
Physikalisches System
Objekt <Kategorie>
Demo <Programm>
Desintegration <Mathematik>
Gesetz <Physik>
Streaming <Kommunikationstechnik>
Negative Zahl
Umkehrung <Mathematik>
Tonnelierter Raum
Metropolitan area network
Kategorie <Mathematik>
Güte der Anpassung
Web log
Motion Capturing
Dienst <Informatik>
Rechter Winkel
Grundsätze ordnungsmäßiger Datenverarbeitung
Klasse <Mathematik>
Abgeschlossene Menge
Open Source
Arithmetische Folge
Prozess <Physik>
Open Source
Statistische Analyse
Elektronische Publikation
Data Dictionary
Körper <Physik>
Komponente <Software>
Abstimmung <Frequenz>
Web log
Formale Sprache
Kartesische Koordinaten
Verdünnung <Bildverarbeitung>
Mixed Reality
Figurierte Zahl
Funktion <Mathematik>
Inklusion <Mathematik>
Zentrische Streckung
Nichtlinearer Operator
Prozess <Informatik>
Installation <Informatik>
Zusammengesetzte Verteilung
Serviceorientierte Architektur
Funktion <Mathematik>
Virtuelle Maschine
Autonomes System
Gebäude <Mathematik>
Kombinatorische Gruppentheorie
Framework <Informatik>
Elektronische Publikation
Dämon <Informatik>
Mapping <Computergraphik>
Bus <Informatik>
Avatar <Informatik>


Formale Metadaten

Titel Designing NRT(NearRealTime) stream processing systems: Using python with Storm and Kafka
Serientitel EuroPython 2014
Teil 35
Anzahl der Teile 120
Autor Modi, Konark
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 10.5446/19962
Herausgeber EuroPython
Erscheinungsjahr 2014
Sprache Englisch
Produktionsort Berlin

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract konarkmodi - Designing NRT(NearRealTime) stream processing systems: Using python with Storm and Kafka The essence of near-real-time stream processing is to compute huge volumes of data as it is received. This talk will focus on creating a pipeline for collecting huge volumes of data using Kafka and processing for near-real time computations using Storm.
Schlagwörter EuroPython Conference
EP 2014
EuroPython 2014

Ähnliche Filme