[OpenLayers-Users] vector layers optimization?

[Deli Soetiawan]

What gis server you use as wms? is it umn mapserver? mapguide? geoserver?

You could precache (render the image first) using Kamap  & KamapCache layer
as a baselayer, Kamap works by creating tiled images in several scale, at
first it might be slow (since it generate many images) but after that tiled
images are cached on server so next time you (user) request the tile it only
send the image without additional process (it don't regenerate image
everytime you request the tile).

Ofcourse you need to able to use .map files as a baselayer inorder to use it
(it would be great if your application is on same server with your gis
server).


Gilles Bassière-2 wrote:
> 
> Hi list,
> 
> In a recent post about hiking trails, I can read: "Be aware though that 
> using vector data in a browser can lead to serious performance problems 
> when the data becomes large."
> 
> Well, I ran into this same problem a few days ago and eventually choose 
> a WMS-based solution. Anyway, I remain very interested in client-side 
> vector rendering. Before exploring further, I wonder if anyone has 
> experience to share regarding large vector dataset.
> 
> In my case, my vector layer featured up to several hundreds geometries 
> (and the user may overlay more vector layers!). Such data volume seems 
> definitely too high for vector rendering but still, I guess it's 
> possible to enhance performance with a clever configuration.
> 
> I can see 3 ways to reduce the load on browser (please feel free to 
> correct if I'm wrong):
> 1) reduce the file size
> 2) use a format that is efficiently parsed
> 3) reduce the data extent
> 
> I attempted to switch my layers format to GeoJSON instead of GML. I 
> expected smaller file size and faster interpretation (given that JSON is 
> native Javascript). Unfortunately, I can't see any real performance 
> improvement. The file size is half as much but my browser still hang for 
> a while (transfer time not taken into account). I confess I haven't been 
> into rigourous tests or preformance benchmark.
> 
> I haven't investigate much about reducing data extent. In both case the 
> whole dataset (full layer extent) is fetched and parsed. Anyway, loading 
> only the visible extent is not useful at small scale and I wonder if 
> it's a good idea at large scale: file size would be dramatically reduced 
> but merging existing data with new data might be too complex, moreover 
> downloading and parsing data on each pan or zoom might break smooth 
> navigation.
> 
> Regarding strict impact of vector data volume on browser capacities, is 
> there a "best configuration"? Any difference between WFS and GeoRSS OL 
> layers? Is there a recommanded way to use GeoJSON (other than 
> Layers.Vector+Format.GeoJSON)?
> 
> Also, I know that MapBuilder used the Sarissa library to deleguate GML 
> parsing to the browser XML engine. As far as I remember, the performance 
> were not too bad. How could this be possible in OL, should I try to wrap 
> the library in an alternate Format.XML?
> 
> Regards,
> Gilles
> 
> -- 
> Gilles Bassiere
> MAKINA CORPUS
> 30 rue des Jeuneurs
> FR-75002 PARIS
> +33 (0) 1 44 82 00 80
> http://www.makina-corpus.com
> 
> 
> begin:vcard
> fn;quoted-printable:Gilles Bassi=C3=A8re
> n;quoted-printable:Bassi=C3=A8re;Gilles
> org:Makina Corpus;GIS
> adr;quoted-printable:;;30 rue des Je=C3=BBneurs;Paris;;FR-75011;France
> email;internet:gilles.bassiere at makina-corpus.com
> title:Web GIS developper
> tel;work:+33 (0) 1 44 82 00 80
> x-mozilla-html:FALSE
> url:http://www.makina-corpus.com
> version:2.1
> end:vcard
> 
> 
> _______________________________________________
> Users mailing list
> Users at openlayers.org
> http://openlayers.org/mailman/listinfo/users
> 
> 

-- 
View this message in context: http://www.nabble.com/vector-layers-optimization--tp19834784p19870688.html
Sent from the OpenLayers Users mailing list archive at Nabble.com.