FeatureLayer load performance

[EdwardSon]

Is there any other performance improvement technique avaiable?
Using OnDemand with OnDemandCacheSize = 1000.
Also setting the MaxAllowableOffset based on map resolution on map resolution changed.
For some of the layers this really improved performance.

Still would like to increase feature retrieval/update performance for some of the layers (particularly those composed of lines).  In particular becuase now have to do update each time the resolution changes.

It seems like the silverlight application builder uses some additional optimization techniques.
If the layer are added using the builder the layers load instantaniously, but they don't exist as featurelayers, but rather as layerinfos in the Current.Map.Layers structure.
Is it possible to leverage additional optimization techiniques.  We need the layers to load instantaneously; it's taking about 6-10 seconds each time the map resolution changes.

[BrentStevener]

The problem most likely isn't related to Silverlight. It's related to your ArcGIS Server dynamic service with a combination of your on-demand caching. 'Instantaneous' performance will only occur if you have pregenerated the cache tiles for your map, while assuming that your web server is running on a capable machine to serve the tiles in a sufficient manner, and that your tiles are created the right way to optimize performance over an internet connection (as in, no PNG32 tiles that are 500kb each in size).

Caching for performance is a large topic, and some places to start:

http://blogs.esri.com/esri/arcgis/2011/03/22/map-caching-tips-and-best-practices-with-arcgis-server-...

http://goingspatial.blogspot.com/2010/11/good-resource-for-esri-arcgis-server.html

[DominiqueBroux]

In particular becuase now have to do update each time the resolution changes.

See this thread for a possible workaround to avoid the Update().

Else as gigemboy mentionned, you'll get the best performances by using a DynamicMapServiceLayer or even better with a TiledMapServiceLayer (images cached at the server side).