Any suggestions on how to best publish 1ft contour lines in a web app? When we publish ours it slows our server down to the point where it is too slow.
Hmmm, one footers would be a massive publishing job. I'd recommend caching it. Here's what Esri says about caching:
The best way to create fast map services is to cache them. When you cache a map service, the server draws the map at a set of scale levels that you define and saves the images. When the server receives a request for a map, it's much quicker to return one of these cached images than to draw the map again.
Here's a tutorial on creating a cached map service:
Tutorial: Creating a cached map service—Documentation | ArcGIS Enterprise
Here's a couple best practices on creating cached services:
Unless you are in flat parts of FL, IL, ND, etc..., one-foot contours are going to be dense, and dense means slow for both the server and client. Honestly though, I can't provide any suggestions or even ask useful questions without more information.
How large of a geographic area are you working with? What scales are you working with? How are the contours being published now, specifically? As Micah points out, are you caching? If so, what are the settings? Are the contours broken up into tiles/grid or do single elevation lines cross the entire dataset?
The first rule of publishing large/complex features: Don't do it.The second rule of publishing large/complex features (experts only): Don't do it, yet.
The first rule of publishing large/complex features: Don't do it.
The second rule of publishing large/complex features (experts only): Don't do it, yet.
Any time I have to publish a large/complex layer, I create a generalized layer to display at reasonable scales above it, and I process the data to display optimally at the scales where it is visible. In the case of contours, this would mean intersecting it with a fishnet polygon, so that individual features only exist in some reasonable distance from the viewfield (at least a 10x5 fishnet, but 20x10 is better), then sorting on the grid cell location (spatially defragmenting the layer) before the polygon ID is discarded. (For points, it would be intersecting by a fishnet, then unioning to make multipoint clusters for overview, and sorting for defragmentation.)
Given a scale-dependent, split, defragmented dataset, if it still draws too slowly, your server is too old/under-powered for continued operation.
Retrieving data ...