In response to Michael's request for requirements:
- All of our points will be read from the enterprise geodatabase (SQL Server), and we would like to cluster multiple layers (6-8 query/feature layers), all containing points. We never used Flex or Silverlight here, however the examples we've seen from both are exactly what we're looking for.
- We are trying to cluster less than 50,000 points, and I see the likely number of points being something more like 500-2,000 points as we're filtering all layers. Some of our filtering could produce points near the 10,000 range, and if we were to fumble and display all the points at once they would number just under 50,000. Our use case is not to make the map faster by reducing points, but to make the points easy/possible to tap on by flaring them out.
- All we need to have happen, is when the user clicks(taps) a cluster of points is for them to spread/spider out. They should then be able to tap a point to see the point's popup window. We have tons of points that share the exact same lat/long, and we can't have points being hidden behind others eg: "oh that business operation wasn't performed because we didn't know anything was there.".
- I wouldn't mind having an optional explode/spider effect on hover as we would use it. However, I can see users with even more points finding an explode on hover feature that cannot be turned off to be problematic.
Example of the cluster flare I've been referring to: ArcGIS API for Silverlight
PS - We also have some maps in the JS API that need this too. extras/ClusterLayer from the JS API does not do anything for us as we have tons of points that share lat/long. If used, the users would have to tap through 20-50 popup windows before getting to the point they need. Luckily for now we're still in our research phase and Google Maps will continue to hum along in prod. But, very slowly ESRI maps are being deployed, and eventually we're going to run out of maps to make that don't need some sort of cluster flare.
EDIT: I believe operationally we could benefit from information gained from heatmaps as well.