[Create]

To be fair, the metric for benchmarking is delivered/recorded luminosity. The machine delivers, and the detector records. It is this efficiency that funding agencies are shown in the periodic reports.

Strictly speaking, (down)time can be "irrelevant", in the sense, that with higher luminosity you get more data (LHC can catchup on Tevatron easy). You can have 5 nine uptime, with 1 bunch circulating, or lots of bunches of particles (the beam is not continuous, it comes in trains of particles). So one thing is, that you also go for as many bunches as possible...

But the periodical reports are on cdsweb, because of the public funding agencies ie. for the machine itself, setting an upper boundary: "Downtime statistics over the 2010 run" -- Chamonix 2011 Workshop on LHC Performance, Chamonix, France, 24 - 28 Jan 2011, pp.70-74. Then the DAQ of your experiment of choice comes on top of this...