Are multi-layer caches worth it?

The requirements for one our projects had eliminated the use of single fused cache for the web application. I was doing some research into whether using a multi-layer cache would help the performance of the application given the extensive requirements that we are trying to tackle. I was trying to find some broad guidelines to justify using multi-layer caches for our application. One of the helpful guys at ESRI pointed me to this link in the ArcGIS 9.3 documentation page (I couldn’t find it in the 9.2 documentation) that does just that. http://webhelp.esri.com/arcgisserver/9.3/dotNet/index.htm#choosing_cache_properties.htm (See the “Advanced options (cache type)” section). I think you need to be in the 9.3 Beta program to access the link. If you don’t have access to the link, here is what you need to know.

If you do use a multi-layer cache, ArcMap is the recommended client for working with the cache. Using a multi-layer cache in a Web application provides little or no advantage over using a non-cached map service. The Zoom Level control that appears with fused caches does not appear when you use a multi-layer cache in a Web Mapping Application.

The issue with the zoom level control is supposed to be fixed in 9.3. But the recommendation on using ‘ArcMap’ as the client still should hold true. So, if you are thinking about using multi-layer caches in your application, keep that in the back of your mind. But this does make me wonder as to how many people are actually using multi-layer caches with any kind of success in their web applications…