These days, there are so many open source content management systems, it's difficult to keep 'em all straight. When I'm evaluating these things, the first thing I look at is how close is the CMS to the content serving? If they're closely bound, then any efforts at scaling one function (say, serving content) has to be considered for it's impact on another (such as, inputting and managing the content). This is why I outright refuse to deal with products such as Vignette. And yet a lot of products seem to follow that same "the appserver is the content repository, serving engine and workflow manager" model that makes Vignette such an awful product.
I understand that Plone is similarly modeled, albeit on top of a Zope app server. A lot of the blogware (such as the one that I'm using now) that masquerade as CMS' also closely couple the data management with the serving of templatized content output.
I say: get with it. It's the 21st century already -- we know better than to mingle our high request rate content with our low request rate administrative interfaces, don't we? The talk I gave at the O'Reilly Open Source Conference in 2000 is posted online and there are loads of other resources out there. And yet people are still ponying up huge mega-buck deals for Interwoven and Vignette, they're rolling their own one-off solutions, they're making compromises by limited their content to blog-style articles and it's sad. The fact that crap that requires an application server closely bound to the CMS functions to deliver the content are still being developed and deployed tells me that the CMS problem is still a wide open issue. It's my opinion that the there's huge uptapped market there.