I think the word for the SKMS is "aspirational", something to aspire to.
If you don't have a SKMS what will you lose? Well, you'll lose information. At times you'll make mistakes that could have been prevented with a SKMS, or have to rediscover information that could have been preserved in a SKMS. Will the mistakes cost more than the SKMS? I doubt it but that is up to the business case - occasionally maybe.
Then there is a second, less-often-asked question. Even if the business case stacks up, we live in a finite world. Is this the best use of limited funds? I would say even less likely than there being a good business case.
An aspirational object gives us an optimal model - something to steer towards, a final objective perhaps never attained but at least providing shape and direction to what we do.
Finally the book does not warn us about this. it does not say "Don't try this at work". No, it (Service Transition) just says "a consistent set of high-quality guidance", "world-class Service Management expertise", "The aim of this publication is to support ST managers and practitioners in their application of ST practices", "This publication supplies answers", "This publication explores industry practices", "document industry best practices". Nothing about "some of this has never been tried before". Section 4.7.5.1 and a few paragraphs elsewhere vaguely allude to SKMS as a planned implementation over time, but nothing about "WARNING: the next bit is a highly speculative future vision".
So, let us beware those who equate SKMS with Configuration Management Database (CMDB) or ITIL V3's superset of that the Configuration Management System (CMS). SKMS is to CMS as Windows Vista is to MS-DOS. No, as Windows 2012 is to MS-DOS.
So SKMS is nice to have, but it is seldom a sensible decision to build it now. It gives us something to work towards so long as we don't see it as a blueprint for CMDB (just as Bill Gates' house gives me something to work towards as I plan my deck).
And as for CMDB, I think my views are clear. Just start at the top of this article and substitute "CMDB" for "SKMS": nice to have, seldom a good business case, almost never best use of funds.