Merken

GeoServer in Production: we do it, here is how!

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
heard this merger here then 1 of the founders reduced solutions uridine on you're going to go there is a bit about how we it's a good use of the interaction of the honors visited the presentation is a summary of 80 once workshop which I didn't find out what's self times if you want to watch something value was going to weak is meant to be quicker the information is available online so of are actually that's table of contents for the workshop and you can look at the workshop on information and will
be you know we are otherwise you can check is spend boundaries and they go there to to the content
missing information and this is the morning information whether it's is the information you need to understand the because and how can it would use of election if you don't know and I understand that it is model and because basically you don't know which is the measure of user so this thing is to understand what these brothers and Italy's is there is more than 170 is it's time box that means we model as when is will be made not always the branches at any time development that might something the name you might want to use development and production unless you really know what to do it for you actually you have no idea what use it's from where the stable the name it's actually a pretty obvious and mundane what is the difference in every branch leads for 6 months and you start as development then it becomes stable than in the government dance and then someone has to pay if you want is because we are not going to that anymore so as to when you see if you start now you might want to look at development for the best thing so you see the new features you wait for development of become stable you does that the little bit and the production you always want to have stable orientations if you follow it if you don't need the officials you can operate once amount and keep using and they'll say OK you several with greater controversial story we usually guarantees that if the job is not too long do so without ontological upgraded configuration if you jump from let's say 2 point so to the 1 8 and the answer now what's the point you might issues because we cannot guarantee that such an older version would upgrade otherwise you can but if you follow the rules you won't be asked what they do with as a user I say Senseval development model to clients and talk to them about the of use of result is extremely nice if you're a developer if you have someone want to mundane reduces the amount of frication actually you don't want to operate and the Gulf actually want to live only when it's needed so and then in the on average if you don't have any problems you can do it once a year and that's fine number scalability is is uh induces we do best and use of quite a lot all logical with unit testing blossom you're testing every nite for the users should be the compliance interoperability testing actually has the compliance with OGC but also because the things that the work in infidel and by sending requests the unit testing is not enough and you can't does that mean American-born the water level you put them together and with stops working that's why you need also known as the best means to stand up due somebody has that send the request to check the results that something we do every nite when we presented activities so you shouldn't be scalable and use United even they come from the right branch if government variants and use it to point 5 right now that is about fixed than it tomorrow next it will be the last thing is that is that's what we do this by before the start the that's talking about tweaks and that within the 1st thing you need to understand is actually you have this little bit about your use of another we use of services and the deployment environment because you can test everything in environment you put in a different environment and they know that is not about you would not be able to replicate the entire environment he might be even the cost of a consistently the securities is that these the and it's a thousand different degrees of freedom the 1st of these issues to user brain and understand what you're going to do not just look for the whole annotation and look at the logs that the objectives spoke to the
mall it actually showing the song it can all of music we want to images of added that the fastest among the states about scalability performance in a lot of it's a it's small part your performance and perceived performance the difference is that this is performance it's in the eye of the use of performance it's something you can measure the performance is not the same thing to give an example of about what the difference is think about time lapse and environments the aside if you think about that you make it out it would take more time the fill them with respect to the entire but that that's the performance it's but why because the user see something right away it it to the effect of a spider but seemed to care about that as well it's gonna be it's a being able to cope with good results and increasing nodes performance is simply the buildings are a problem you can be extremely fast not scalable at all I argue that you give all the available resources that's in the west and you process that it was the time so you have fast but you won't scale you can be an extremely scalable and extremely small you try to use as less so as possible and put together as many requests as possible of course you want to be the fastest available robustness you don't break and half of that is a long list
of different things you can do I will not go into details of all of all of the example I would not explain a lot of although you can prepare you to arrest input vector because we'll been talking about these 4 images that is above the augmentation a basically the
information is you know is that you need to understand your use cases you need to understand what the user wants and the environment many people that most of the time discovery away experimentally format conversions compression without without any clear in mind what they're doing and give an example if you want to optimize access to necessary and read if you don't know anything about the and you want to optimize blindly the best thing to do is to not do anything OK why because usually this formats their use in environments where the data is relatively small in dental sponsored grades so the use of it pretty fast you don't want to spend a lot of time and preprocessing the 1 key point to remember when you proposes that is that it takes time of day especially for the larger so all used start doing sophisticated preprocessing might give you best performance but in certain cases my makes the old which means used to think about study come all the metallurgical model models which ographic models the key is to actually get it out as quickly as possible so will be talking about this and some of the presentation of example yesterday ages and we're done many tweaks in order to although you to sophisticated things like window roles concluding except except on the fly without the preprocessing because the key was to get it out as as we can also assume we want to look performance be careful with the pharmacy you use some of them may not be a good fit for performance if your if your format is an ASCII format the money from the logic I go quickly about the 1 thing I can tell you we plan to use what we do sophisticated preprocessing we can always use student Judith is like this we slide we can support many different optimization you can use of the rate limit of 4 gigabytes when you have extensive that you might want to use things like mosaics or a or we could talk for hours about what to use when and there is no single magic recipe and that from it is all under talked about we're actually moving polewards making it easier to put updates large periods that says there are more about attitude toward data on top of existing that that without being to exact and the older but if you want to talk about you can read some of the of and use of that would not be so they can be a lots if you can exploit find them ourselves and create a special
temporal period so you know it's easier to do updates and and other the lecture because things
and you really want to use in the manager has enabled format and if if we need to explain why you might need to use a little bit of title studies but another is usually want use of ontologies for vector that are easy and started fast when you're going to extract a small portion of that most of the time although it might seem like that much of it is a structure to these logical records and what you want to build usually is extract create a portion of this is is it is that's it this can be of fundamental or can be spotted on a combination the weekly shared by school as long as you don't need complex if you need to run sophisticated and complex queries using often America attributes messages over there is no support for it in shape files on the indexing of semantic if you have a 4 gigabytes safe that you want to use as background layer and you have a complex side eventually style things differently filtering them depending on on the of phenomena that use and so that think about you want to put this this concept of because otherwise you said that every time will then dived deficit filtering is Shafer we all the special I know that might seasonal you think distance but it that I can read the ones with a client because the people at the tendency to put everything on the screen and to around original poster and of course it is not the best thing to so Schafer is good if you are actually know you're going to end everything and you're only going to feel that they can talk when sponsored 15 and at that point shape are usually most possible especially the basis because the number 1 problem of special base and then just the scalability not performances and what that is that you have meaning that in the number of of connections that you can use so you can make enquiries as fast as you want problem and you want that 20 thousand connections of so let's say that's where the leading dishes on from think about for example a simple map Brenda from posterior follicle let's say wouldn't and of these maps in 15 ms so we can know how many requests that's trying to plant plant 20 OK so that means if you want to go to and this was the 2nd you need at least 20 connections In most use cases in enterprise applications you don't that that many connections because that would be the share of you know you don't control of them so you would need to take this and about scalability will be initially you might be fast must of it will be relations initial so you will have to use your connection wisely hinges on the crime if I'm wrong with tried you still answer for them and that it 1 for reading data 1 4 and that why we could load everything in memory and for that connection away but that would be will give us feed on calibrated remember what was said before because it will put all the resources on request so we don't go below this thing will child so well relating with fast with scalable the downside is that we use the collection for the longest time we don't use the convention for the entire time of the request policy 60 per cent 70 per cent you it depends on that's where so that gives India if you know all of us to all many connections you need for set trouble but you want to reach that of course you need to be indexed you know that if you know how again you might want to read a little bit of special databases works but if you need out if you and I remember if it's raising or lowering the debugging and you want to look all the most of something like that just that was speak this sequence where is that it's actually send it to the best of its so you can get for them in the event of its full and analyzed duties and ultimately there properly handle connection point in an enterprise environment in my it might seem simple and it is not that simple depending on how our environment destruction if there is a the period in the midst of actually killing connections and you don't know what there might be softer than other appliances that keen interest in the conventions when the arrival we learned that the hard way so you really want to understand how the quantity of connections indeed it's not all about validating you always want to validate the ones that are also wasteful actually in background you make sure that the body and then there's the usability because it does not at the very least we get answer in about the case you would do stuff gets and because of their connection board exhaust
week suggested that it's water for shapefile and special that arises when you have many many attributes in your tables in your book not going to use them in a style in your except that you might want to leave them aside and then that will make the ship was called body will also make use of the use less memory for the special but this connection because well was such that that will freshness just should be is not enough for basis for not load article is that it doesn't matter but it has a so the 1st of the optimization is to use your brain and so this is an easy optimization with the parameter of interest OK you to him in so
stylized while this has many citing capabilities you wanted to make maps of readable by the user so never tried shorter much of the like you if you have a lot of the federal network of a nation you don't want to show it completely when you are the 1 1 medium so again because it would be a black they say what it's wasting to use any itself with in the time of the users because they won't be able to to
them but mentioned to labeling labeling is also expensive because there is a conflict of solution engine going and it's gonna do some of them school place the labels right so around we have a new optimization in terms of quality in just over 2 9 which makes for better special occasion because we allocate space for chart as opposed to the volume of of labels bodies having some over and so there is a system variable to turn it off if you if you wish if you find that there it's slowing down your mother production to much avoid using
too many features such as because user will allocate multiple renderings of faces to so you'll answer your request which means it's gonna use more memory
and so if you're using as an ordering which is a new about some of the data that we have in danger so it's 9 all 2 will you know make sure that underpasses and overpasses and stuff like that are in the right position in the men also be careful about in not doing that articles levels because we have to keep a certain amount of data in memory to do to go back and forth and and do this all the z of income
alternate caching wages with receive piling is always very important thing to consider when we're doing maps because everything that's not changing can be cashed uh provide a very significant speedup so play now what players you wanted to connection which you don't because white and
you want to talk as well because they can just take a lot of disk space is so you might not have enough for everything so duties and medallions user so that user us speed was compared to our an external such leased
space considerations so take into account the car going up to level to any might require make a bicycle not terrabytes of storage so depending on your area of course so they can take that into
account take lot take into account the client-side caching sometimes you can tell the thought the client and not to request the same time time over and over for us in the amount of time and that's sort of the ultimate ultimate optimization because the client 1 the dissimilar told it's not just not producing the time is not even transferring which is great
I choose the right format for your tireless so relates the PNG format top NJ back for the rest of the time we have a then you inject were PNG if you're amazing have transparency so consider using that there's also vector it's a community model but it's a very nice and to recognize more compact they uh and they can be over resumed so instead of giving your vibration level 20 you can just say talk about I don't know 16 17 and over zoom in on the client side because this information that I just over 0 2 9 we haven't got support which is very nice to the people around tigations between servers so you can actually prepare action once a offline and then put it on the production server which is again a nice for you user good experience our mn could see their of choosing between uh when you have a cluster when you have a cluster you might want to choose between having a single share of the cash which is probably the best option if you will are proceeding everything versus more independent caches local to the hard right uh when you are using this time just catching on demand the the figure here is that not having to write all that much on a shared file system because most cheerful a lot when you're here when you people writing and create a new tires and the deleting times continuously of
similarly here with the 1 thing about this you can actually
mix-and-match independent gesture should attaches it was the 1st layers can use different storage so you might want to use sugar caches for layered
unchangeable like background smaller caches the changes more frequently the leaves will take forever if you put them on a shared ownership of things and work so I show system so you might want to consider that in them independent which means duplicating the they apparently doing that but if you compare the time it takes to the need created then you actually much
faster OK the assume we optimize everything you want to measure what did not and this is under explaining why go it's your friend literature look and the results it's still courts I mean engineer another so and the simplified if you're not that they would with the calls and all things editable context but that like much good and if there is if there incurred the 1 but see almost literally neighbors and the response time and the bad red 1 is the resource optimization systems again it's a simplification after the water not OK they can't go the it this way each when you have toward them out there might be fluid which and there will always be some preparation but where simplified what do you want to do when you optimize it's actually having the could girls it's been rolled usually it's measured by increasing user of France so you want the throughput the world more request that 2nd going and you on the response school to stay down as much as possible and this at point you would need the Baltic that it can be this CPU it can be reduced if you can be done took it can be the melody can deduce because I mean the bottlenecks in the software as well OK that is 1 that we talked about before and there is some mentioned so then the object is actually when you measure it still improve the schools we see once you optimize over that you want to make sure that the throughput will not fall down OK as you reach a maximum efficient but you will ensure fairness so if you get more requests and that was it you can cope with it was that during the oviduct and that is so that the Serbian all of them and
not being able to set and because
of Russian universities in the mold keeps increasing up to a certain point these were not about that we get words so we need to make sure we don't reach that situation that is the waste the Buddha is served in Greece situation 1 is to send the request to big and we have the resources to manage their services blocks and make sure that 1 the other ways to send legal requests listening to my neighbors and I'm like a denial service of each service as parameters and settings
actually make sure you get linear almost remedy we use all minerals with it all much data we generate and for longer service can
run and this depends on the service that is set this problem as the defense of assessing there will be a sum that showed some announcement of the appears once you can make sure that this process standard you for do much about music you do much it's not as simple as that the processes to cooperate because you cannot keep that I've been jealous want kinda than that application that is a controversial but you know you can install to make sure that you do quality of service and product link which means if I get more requests and something and I was started you the requests instead of trying to serve them and I can actually also audio quality of service like think these users and I want to my true but cannot be more than exit was and that is the time to explain all that works but it's because of sophisticated
and it actually goes into look something like this and the blue line it's actually it's a measure throughput so you see at a certain time is that the world so what you usually do use condoms to make sure that when the you put example in this case you certainly not execute more than 60 everquest at the same time the for not 60 request a 2nd if it's a request executing 15 minutes it will be 16 that plant that was it is that the request that executing if you for courts it will make what and expand its body for costs idea you won't be able to execute more than 8 thousand requests at the same time it's not at the same time like a line of code about executing there will be some idea awake so if if for or it could be a because they could switch from the right except but they will never be able to OK OK that there would be an
interesting but again against I
do that time and if you check the
presentation it's about what to do about the what's when on in the park so that is 1 thing which I
just want to show you the you need to make sure when somebody's production when the faster the fact I had another thing
modeling and whether this you need to be prepared because everything with things made if it doesn't know what it is in is so keep that in mind you just need to be prepared for 1 so monitoring blogging and the date is naturally uh method and its effects that's it and that
yeah so maybe 1 quick question because I was a little later anyone that maybe you to really the beginning is says that this there's no stable release and morally Anderson's correctly and there was no stable release 4 which is some loss there's always the warriors with this slope
maintenance or I just wasn't what is it exactly at the same time every time there is the development branch a stable branch among variants of the releases of 5 also so you get a real users of Islamist while flows and select and discover so we only use a 1 month of the stable woman naming the maintenance next
month at the end of the woman and so on and every 6 months we switch the development of a stable the stable but the maintenance and the old maintenance
goes away and then I'll probably adjusted
Server
Bit
Interaktives Fernsehen
Räumliche Anordnung
Information
Inhalt <Mathematik>
Kombinatorische Gruppentheorie
Computeranimation
Tabelle <Informatik>
Resultante
Bitmap-Graphik
Bit
Punkt
Komponententest
Versionsverwaltung
Familie <Mathematik>
Bildverarbeitung
Information
Login
Computeranimation
Metropolitan area network
Client
Skalierbarkeit
Geschlossenes System
Prozess <Informatik>
Softwarewartung
Einflussgröße
Kontrolltheorie
Softwaretest
Geschlossenes System
Computersicherheit
Java Enterprise
Applet
Biprodukt
Funktion <Mathematik>
Rechter Winkel
Server
Vier
Fünf
Information
Datenfusion
Orientierung <Mathematik>
Subtraktion
Server
Quader
Zahlenbereich
Content <Internet>
Vektorraum
Freiheitsgrad
Inhalt <Mathematik>
Modelltheorie
Softwareentwickler
Konfigurationsraum
Binärdaten
Verzweigendes Programm
Schlussregel
Objekt <Kategorie>
Räumliche Anordnung
Modelltheorie
Visualisierung
Beobachtungsstudie
Bitmap-Graphik
Soundverarbeitung
Resultante
Zentrische Streckung
Subtraktion
Spider <Programm>
Gebäude <Mathematik>
Varianz
Mailing-Liste
Vektorraum
Ein-Ausgabe
Computeranimation
Metropolitan area network
Erweiterte Realität <Informatik>
Knotenmenge
Skalierbarkeit
Geschlossenes System
Gruppe <Mathematik>
Ein-Ausgabe
Mereologie
Kontrollstruktur
Bildgebendes Verfahren
Aggregatzustand
Bitmap-Graphik
Offene Menge
Stellenring
Umsetzung <Informatik>
Konfiguration <Informatik>
Punkt
Minimierung
t-Test
Parser
Euler-Winkel
Computeranimation
Gradient
Dedekind-Schnitt
Bit
Geschlossenes System
Parkettierung
Bildschirmfenster
System-on-Chip
Quellencodierung
Geschlossenes System
URN
Präprozessor
Simulated annealing
Mosaicing <Bildverarbeitung>
Güte der Anpassung
Quellencodierung
Bitrate
Frequenz
Dateiformat
Datenstruktur
Zahlenbereich
Ein-Ausgabe
Dateiformat
Information
Ordnung <Mathematik>
Warping
Schlüsselverwaltung
Zentraleinheit
Fitnessfunktion
Server
Teilmenge
Kombinatorische Gruppentheorie
Mathematische Logik
ROM <Informatik>
Checkliste
ASCII
Mosaicing <Bildverarbeitung>
Inverser Limes
Modelltheorie
Neunzehn
Beobachtungsstudie
Formale Grammatik
Solitärspiel
Elektronische Publikation
Konfigurationsraum
Einfache Genauigkeit
Bildauflösung
Einfache Genauigkeit
Netzwerktopologie
Komplex <Algebra>
Dimensionsanalyse
Räumliche Anordnung
Modelltheorie
Mini-Disc
Innerer Punkt
Manufacturing Execution System
Retrievalsprache
Bit
Punkt
Gemeinsamer Speicher
Summengleichung
Parser
Orakel <Informatik>
Kartesische Koordinaten
Textur-Mapping
Computeranimation
Eins
Formale Semantik
Client
Dedekind-Schnitt
Skalierbarkeit
Geschlossenes System
LASER <Mikrocomputer>
Eigentliche Abbildung
A-posteriori-Wahrscheinlichkeit
Automatische Indexierung
Shape <Informatik>
Kontrolltheorie
Benutzerfreundlichkeit
Datenhaltung
Abfrage
Aliasing
Dateiformat
Frequenz
Ereignishorizont
Quader
Datenstruktur
Einheit <Mathematik>
Festspeicher
Zahlenbereich
Datenparallelität
Lesen <Datenverarbeitung>
Message-Passing
Server
Folge <Mathematik>
Teilmenge
Schaltnetz
Vektorraum
Zahlenbereich
ROM <Informatik>
Whiteboard
Datenhaltung
Checkliste
Datensatz
Verzeichnisdienst
Datennetz
Abstand
Datenstruktur
Touchscreen
Tabelle <Informatik>
Einfach zusammenhängender Raum
Beobachtungsstudie
Binärdaten
Elektronische Publikation
Ontologie <Wissensverarbeitung>
Konfigurationsraum
Vektorraum
Elektronische Publikation
Mapping <Computergraphik>
Last
Komplex <Algebra>
Parametersystem
Identitätsverwaltung
Basisvektor
Dimensionsanalyse
Attributierte Grammatik
Unternehmensarchitektur
Shape <Informatik>
Einfach zusammenhängender Raum
Parametersystem
Schlüsselverwaltung
Datennetz
Wasserdampftafel
Minimierung
Thumbnail
Extrempunkt
Rendering
Computeranimation
Schlussregel
Maßstab
Last
Festspeicher
Basisvektor
Mapping <Computergraphik>
Attributierte Grammatik
Tabelle <Informatik>
Objekt <Kategorie>
Betriebsmittelverwaltung
Server
Raum-Zeit
Minimierung
Flächentheorie
Versionsverwaltung
Varianz
Biprodukt
ROM <Informatik>
Rendering
Term
Raum-Zeit
Computeranimation
Gesetz <Physik>
Multiplikation
Geschlossenes System
Festspeicher
Volumenvisualisierung
Räumliche Anordnung
Spezifisches Volumen
Versionsverwaltung
Betriebsmittelverwaltung
Caching
Einfach zusammenhängender Raum
Retrievalsprache
Automatische Indexierung
Konfiguration <Informatik>
Datentyp
Adressierung
Ortsoperator
PASS <Programm>
Übergang
Dateiformat
Computeranimation
Übergang
Mapping <Computergraphik>
Open Source
Festspeicher
Caching
Parametersystem
Zoom
Protokoll <Datenverarbeitungssystem>
Äußere Algebra eines Moduls
System Dynamics
Ordnung <Mathematik>
Rohdaten
Raum-Zeit
Übergang
Raum-Zeit
Lie-Gruppe
Menge
Computeranimation
Flächeninhalt
COM
Mini-Disc
Zählen
Zoom
Speicher <Informatik>
Mini-Disc
Meta-Tag
Bitmap-Graphik
Stellenring
Gemeinsamer Speicher
Minimierung
Template
Computeranimation
Übergang
Spezialrechner
Client
Rechter Winkel
Total <Mathematik>
Dateiverwaltung
Figurierte Zahl
Caching
Teilbarkeit
Güte der Anpassung
Ähnlichkeitsgeometrie
Übergang
Programmierumgebung
Zoom
Biprodukt
Dateiformat
Erschütterung
Konfiguration <Informatik>
Web log
Rechter Winkel
ATM
Client
Server
Dateiformat
Information
Decodierung
Gruppenoperation
Vektorraum
ROM <Informatik>
Systemprogrammierung
Datennetz
Zoom
Modelltheorie
Gruppoid
Operations Research
Stochastische Abhängigkeit
Modul
Solitärspiel
Elektronische Publikation
Raum-Zeit
Konfigurationsraum
Browser
Vektorraum
Quick-Sort
Caching
Mini-Disc
Caching
Stellenring
Elektronische Publikation
Gemeinsamer Speicher
Konfigurationsraum
Mathematisierung
Programmierumgebung
Computeranimation
Endogene Variable
Systemprogrammierung
Last
Geschlossenes System
Datennetz
Caching
Speicher <Informatik>
Operations Research
Mini-Disc
Stochastische Abhängigkeit
Resultante
Punkt
Wasserdampftafel
Minimierung
Mathematisierung
Systemaufruf
Zentraleinheit
Kontextbezogenes System
Computeranimation
Objekt <Kategorie>
Metropolitan area network
Geschlossenes System
Software
Endogene Variable
Response-Zeit
Kontrolltheorie
Parametersystem
Geschlossenes System
Fehlermeldung
Server
Punkt
Verhandlungs-Informationssystem
p-Block
Computeranimation
Inverser Limes
Menge
Server
Wort <Informatik>
Grundraum
Prozess <Physik>
Gewichtete Summe
Gewichtete Summe
Kartesische Koordinaten
Information
CAM
Biprodukt
Binder <Informatik>
ROM <Informatik>
Code
Computeranimation
Warteschlange
Metropolitan area network
System F
Gruppe <Mathematik>
Server
Kontrollfluss
Dienstgüte
Parallele Schnittstelle
Kontrolltheorie
Gerade
Einflussgröße
Standardabweichung
Server
Last
Maßstab
Vererbungshierarchie
Konfigurationsraum
Räumliche Anordnung
Orakel <Informatik>
Verbandstheorie
Biprodukt
Kombinatorische Gruppentheorie
Systemaufruf
Computeranimation
Soundverarbeitung
Server
Einfügungsdämpfung
Räumliche Anordnung
Computeranimation
Mittelwert
Softwarewartung
Stabilitätstheorie <Logik>
Reelle Zahl
Trennschärfe <Statistik>
Besprechung/Interview
Verzweigendes Programm
Vorlesung/Konferenz
Web Site
Softwareentwickler
Datenfluss
Besprechung/Interview
Computeranimation

Metadaten

Formale Metadaten

Titel GeoServer in Production: we do it, here is how!
Serientitel FOSS4G Bonn 2016
Teil 69
Anzahl der Teile 193
Autor Giannecchini, Simone (GeoSolutions Founder)
Aime, Andrea (GeoSolutions)
Lizenz CC-Namensnennung 3.0 Deutschland:
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 10.5446/20401
Herausgeber FOSS4G
OSGeo
Erscheinungsjahr 2016
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract The presentation will describe how to setup a production system based on GeoServer from the points of view of performance, availability and security. The suggestions will start covering how a single node GeoServer should be prepared for internet usage, tuning logging, connection pools, security, data and JVM preparation, keeping disk, memory and CPU usage in check within the limits of the available resources. We’ll then move to tools used to monitor the production instances, ranging from probes to request auditing and watch-dogs. Finally the presentation will cover setting up a cluster of server and the strategies for keeping them in synch, from the traditional multi-tier setup (testing vs production) to the systems that need to keep an ever evolving catalog of layers constantly on-line and in synch.

Zugehöriges Material

Ähnliche Filme

Loading...