Well yes but AFAIR MS at least provided some data that could be critically analyzed (whether it stood up to the analysis is a different matter.) I am all for using TCO as an argument in tech advocacy, just as long as it has some backup in verifiable data.
A discussion on TCO is very dependent on the context in which the analysis was made, such as application type, volume of data stored, volume of data indexed, type of data (e.g. is it a tree? is it a graph?), queries per second/hour, number of concurrent connections, analysis on further scaling possible (vertical / horizontal), analysis on the requirements for real-time ad-hoc queries, entry-level costs, and so on and so forth ...
Basically, the variables are so many and so complex that any TCO analysis (regardless of validity) won't apply to your case.
That said - I rarely feel the need for NoSQL, simply because I can fine-tune PostgreSQL to whatever needs I want. But I'm not running neither Facebook, nor Twitter.
Also include the skills of the people operating everything. A MCSE trying to administer various Linux boxes will have terrible TCO, and a RHCE trying to optimize an NT box will have similarly bad TCO.
Comments
Well yes but AFAIR MS at least provided some data that could be critically analyzed (whether it stood up to the analysis is a different matter.) I am all for using TCO as an argument in tech advocacy, just as long as it has some backup in verifiable data.
Basically, the variables are so many and so complex that any TCO analysis (regardless of validity) won't apply to your case.
That said - I rarely feel the need for NoSQL, simply because I can fine-tune PostgreSQL to whatever needs I want. But I'm not running neither Facebook, nor Twitter.
Also include the skills of the people operating everything. A MCSE trying to administer various Linux boxes will have terrible TCO, and a RHCE trying to optimize an NT box will have similarly bad TCO.