Add to Watchlist

MapServer MapCache: Project Status Report

14 views

Citation of segment
Embed Code
Purchasing a DVD Cite video

Formal Metadata

Title MapServer MapCache: Project Status Report
Title of Series FOSS4G Bonn 2016
Part Number 76
Number of Parts 193
Author Bonfort, Thomas (Terriscope)
License CC Attribution 3.0 Germany:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
DOI 10.5446/20412
Publisher FOSS4G, Open Source Geospatial Foundation (OSGeo)
Release Date 2016
Language English

Content Metadata

Subject Area Computer Science
Abstract This talk will present the recent developments that happened in MapCache, the tiling server from the MapServer project. Main features include enhanced support for dimensions allowing easy creation of imagery mosaics, and further integration with cloud based infrastructure.
Series
Annotations
Transcript
Loading...
it I can get a continue now
made them map service sessions and Thomas all for again is going to be talking about them not cash status report Thomas Hi again so covers again and others looking about not cash
which 1 is that tying them to maps of sorts and the next and brand and government government and it's a library used for creating map tiles with client interfaces that's can either be an Apache modules for setting the other 2 from the Apache process and Unix modules of CGI procedure and also receiving it all go through that the main features and it has a different set cash back ends that supported and then in the 2nd step of over the and the new function it is coming in for the for the communities as so it's rather versatile library with that supports multiple cash ends multiple client protocols so that would be in the WGS theorists WMS xyz with and it has time management features that that allow you to pre-generated styles depending on given rules we compress images for a better image compression performances interpolating dials that present that are on the cash that as well there was a no and it's an S C library written in native code so hopefully you relatively fast and supports tumors which is the lowest you'll and training protocols WGS which is the oldest you want but you quot keys which MacGuide syntax and plane xyz on there is also supports WMS request full put support requests this answer Chile time requests from any can assemble tiles from its back and gaseous so it can either assembled them vertically so if you asked for the blending of 2 layers but can itself it returns a single image those 2 2 layers banded together it's also do horizontal and something which means its response to I get up request of the sides that's not the size of the actual time it
also can act as a OGC proxies which means that it's all intercept requests that can be responded to by the configured catches so basically TMS story requests with and then can forward other types of requests to back and real the season was In used to provide a single points to speed up some variants requests and on the same endpoint that over to about the main WMS
but the bucket list of miscellaneous features that can be found on most most of catching engines this is the idea of for caching although I think expiring of all dials on fatalities although they're configured titled and it'll be requested from the backend and result for given number of seconds days without the reporting can be done either by our standards that are essential in the image philologist status code depending on what kind of information you want to send back to the users supports meta filing when you're creating images watermarking is you need to have role to reach titles and then next question which is of sampling audible titles so if you're arrested data for example oscillators only the president up to zoom level 14 and it respond to the higher level time requests by sampling the the highest levels of data that you actually have but now operations that span can be done on the images returned from the river so as before storing them in the cache is basically choosing what this format to use before sorry the cash but it would avoid double data compression so you request GNG data from the US from you know WMS and and then just go with the lossy deeper compression once you saw it you can also choose to be very aggressive compressions and quantity you to apply to png is when you are creating data so heavy is that the compression of all you can choose a shown rendering tiles on the fly need to due CPU lights compression to to choose a low compression levels on requests in the image formats that supported on dingy which you can choose the compression level and you can also decide to apply quantization so reduce the 24 bits PNG to work a bit but it been G to achieve that you can configure them color model used at universities here and the Congressional and there's also automatically mix the energy breakfast PNG format very useful for us today so that doesn't have full coverage so depending on the actual date of that thiomersal selects either uh due compression which applies for images with their which don't have any transparency and for the areas that on the border where you would need to support transparency that in that case it's also PNG yesterday from storage there's a detection of MT uniform tiles which tend to happen quite often with the fact they depending on the Kashmir Kansas will either be done by a symbolically linking files along and just storing and quadruped in their cache and then that energy reconstructing the 1 bit Kinji from that stored color or you can just I that your your cash for the CD than that that's entirely should be treated as being fully transparent I given guys can be stored long multiple grids if panels through aliasing so basically the and non-standard 990 that encode that was used for not at all can be used alongside the official that uses 57 1 and that control we have to use 1 cash for that innocent until and that of schools are equivalent and can be sold in the same cash we can also configured agreed that subset saying you only want to cash certain zoom levels so that interest you or restrict the extends of which a caching only for a given country and if you need to work expand it'll come with extents for some reason or the other then you wouldn't have to receive only caches issued in so the caches themselves are aware the actual time that the the 1st conceptually it's very simple it's a back and it's able to store the difference data brother for a given y z once you implement the exists gets put in the levels and and that becomes a candidate for becoming a Catholic for cash and there are some specific texts that used to some time for that the detection depending on the tone available and certain gas records discussions on the really
simple but the titles on the file system mainly use of 5 so for testing purposes or for limited datasets 5 systems tend to have issues with the when storing a very large number of thousand and then you get issues once you try to manage those data listing the size of the cache is transforming them from 1 case to the other what and you can either use configured there structure all you can also give it a template to reuse existing discussions you can still and
inside the rescue the files so a single contained file that contains all you can split those caches into multiple database files and if you need to do use different files for different groups of suppliers sets dimensions configure the maximum number of styles that you want per database files so the big advantages single file and the manageability of its own supports India iTunes schema that's you you could have as previously created and it's rather efficient in the disk space so the columns of very large caches kinds can get back a bit complicated to to manage and concurrent insertions are slow inside the skewness of skew it will be looking the whole database when is right on its own just an
example creating a secure providing them under actual SQL statements that are going to be run by cash to extract the files for given explain the well supported back
and is the Memcached instances which is ideal for was the temporary data that you can we calculate so that becomes stable after a few few days so a huge advantage is that it's a it's a speedy you can easily distributed and have a single non-cash instances instance serving multiple but services is automatically prune your data so the least recently used data will be automatically expired and removed from you and of course the limitation is that you are limited by the available grammar on that services limited storage we also supports the http URIs caches so that's storing gaseous true the basic rest GET PUT and delete operations the the external entities piece you can either do purists and we also provide authentication and authorization plug-ins for hitting the other child providers for it and that the object storage so either with history through through your there their key authentication and that much suffers 0 good origin forestry and just solution can can also use the the temporary keys that are provided to you by the and the Cloud instances instances metadata so don't have to actually publish your your keys to the 2 configuration files there's 1 kind of cash which is use essentially for the rest of data so it's actually using Due files with the internally time structure and reading that data directly from that used to finance and that means you can start using a disk based so caches again because the number of actual files restoring is much lower has it's also means that those 2 files can be used outside of connection can be used in any GIS applications mission to provide them for another WSN theory predictions for example for whatever you use you probably have to set up you know all Rendering trained for creating those duties because fly creation of those has issues due to the the format itself basically that's you can only have 1 rises so that the final and that you can't really updates data inside a given defined you can actually do it but you won't be reclaiming the storage space of the updated data on the 2 data and of course is that this specification and it only supposed to take image so if you need to handle transparency that's not the we have other kinds of
meta caches composite caches so you can combine gestures with different performances options and the 1st type is called the composite cash so basically depending on some rules for a given time different actions will be selected for the the time being the rules you can apply depending on the zoom level you can have 2 different gaseous and depending on the British depending on the dimensions of of of your data is an example of using 4 caches from zoom 0 2 8 of restoring a discussion I wanted to be but data and then for looms over 9 also written and catch data that be bottom you temporary they we'll sort of
multi-tier captures the multi-tier caches allow have the best of both worlds between cheapening that slow and large caches and fast and inexpensive and and small cages so basically if conditional and find out and tile in 1 of the 1st caches will then go back to the slower cash but that that's the data pull it up put it in the in the past the cash flows subsequent Texas and then return and
this is another list of supported caches which I won't go into into detail then you can start a thousand distributed the react tested in but KDD files and then and may be coming soon in the country's factions in the and I we provide a pseudo which allows you to pre-generated titles so they're not trend on the 5 it's pre-generated beforehand it's a multithreaded on multiprocessors during it's doing this using jobs in parallel you can use this either for different scenarios i that seeding all-cash if you want to which you can also choose to selectively seed subsets of what's in what interests you so interest to see crossing a single given dimensions you can regenerate files that are older than the given date so that you can restrict to the result you want to regenerate so generated and all the extent we can also use geometries to only generate generates other tiles that intersect the geometry that you give it to some examples of what
you can do with seating so 1st example is just saying futurism level 0 to 12 2nd example is using a should file containing countries and say OK I only need to seeds level so seems to 16 where I actually have data and don't see the world as the and you can also do it for long region automatically induce regenerating trials this year change that geometries In the example here connects to an external pushes database that the posters databases through G and will receive the titles that intersects the geometry of a given change that we support the
concept of dimensions activity that would be time filter or elevation it is very useful for meet data are at issue of different styles for different kind ideas of different applications will post consists of can be used as the dimensions so you configure and provide a single file but then the kind and selects a subset of the tacit giving the nation values that's that interesting so the time dimension an example where the time dimension would be to create a mosaic of individual satellite images and you want to create a mosaic of all the images that were created for a given month for a given day or given you they also a mosaic
the mentioned it can be used outside of time requests and building the mosaic based on business rules of association of multiple images and you want to create a single mosaic from those images that can provide a database of the available subimages you have and then creates a mosaic for each given dialog it's a list of corresponding images for that tells you the example here is selecting the visual images and ordering them by the highest-resolution images 1st and then by their acquisition data the most recent roundup both those dimensions things have some kind of not stacking because those requests can possibly return you a higher number of that individual subband so it'll stop accumulating cream there that caches and once the idea entirely useful so fully a
so now for the coming and then that's that's the dimensions of his upcoming so we know in the coming not carefully 2nd big upcoming future support for given all sources instead of only supporting WMS services back and sources you can give catch any G dealt supported data sources a use case for 1 of that's 1 would be to title a bunch of images that you only know y dimensions the the 250 individuals In this example the adipic 2005 and get that title to by the cash you can use it for every projecting retiring third-party tiles given that GDL supports dataset was with its so many drivers you can give it to me and that's also supports re-projection on the fly so you that's in 1 way to migrate from 1 context to another griddle or projection and that's
it for this presentation issue have any questions of being please to answer try to answer thanks very
much Thomas for perfect timing and I thought of everything else 3 have any questions for what effect sizes so start again in 10 minutes or so
Lecture/Conference
Computer animation
Computer animation
Computer animation
Computer animation
Computer animation
Computer animation
Computer animation
Computer animation
Computer animation
Loading...
Feedback

Timings

  288 ms - page object

Version

AV-Portal 3.8.2 (0bb840d79881f4e1b2f2d6f66c37060441d4bb2e)