ZEO sounds fabulous, but it may take a bit of time to convince the Deci$ion Maker$ that this "free" system is now worth the inve$tment to make it reliable too. The concept of raising performance with multiple "read only" ZODB servers, all linked to a common backend RDB can be meshed with other ideas that have passed through the Zope lists. One is to use conditional links pointing to semi-static documents in a file system, such as daily business reports, or user defined ad hoc queries. If an URL maps to an existing file, it is served directly. On the other hand, a "file not found" branches to Zope, which returns the update to the requestor, where it will be "cached" as a static file. Yet another approach is to have hard links from within Zope "pages" pointing out to a file based storage. For example, an image repository can sometimes (always?) provide faster response than Zope file objects. A graphics repository, for example, could reside on a common backend optimized for serving shared files. I can't recall any other hybrid strategies just now, but I'm sure others can point the way. Let us know how your experiment pans out, we all need as many good ideas as we can get our hands on!