It is important to distinguish between two fundamental types of Web mapping: dynamic Web maps and cached Web maps.

1. Dynamic Web maps are dynamically generated from a database (e.g., which is similar to multiple map layers in ArcMap).
2. Cached Web maps are authored and processed into a multi scale raster cache. There are several benefits to using cached maps including very fast performance, smooth pan and zoom, and high scalability (almost no CPU resources are used). This type of Web map is used by consumer mapping sites (i.e., Google and Microsoft).

ESRI created a process for creating cached maps so users can enjoy the benefits of this very high performance and scalable environment.

There is, however, some work associated with creating these raster caches. Users must learn the caching process and how to design a good map at multiple scales. Our experience has been that it takes time experimenting and learning how this process works. At 9.3 there are more streamlined workflow processes for creating caches and the documentation describing these processes is much improved. Also at 9.3, ESRI has dramatically improved performance of hybrid combinations of these map services (i.e., cache for maps that don’t change and dynamic services for overlays that do).

ESRI is also actively developing a much faster dynamic rendering engine for ArcGIS Server that will be out in a service pack later this year.