Fortunately, the GROS approach has most often been a so-called "soft launch", i.e. you don't hit all the possible media on the morning of the day of the availability of the new data, including multiple appearances on breakfast TV programmes.
Instead, it's there one or more days previously, and the word soon gets around, so that you don't get the "(English) 1901 effect", in the first few hours, of hunners o' thoosans of folk logging on to find their granny.
In addition, it's not all that difficult to arrange that either (a) there's some extra site capacity in the first few days, and/or (b) when the level of site hits exceeds a predetermined level, arranging for there to be a message along the lines of "the site is very busy at the moment, please try again later" ..... without the whole site crashing.
Did anyone ever come across an explanation of why the English 1901 went offline for weeks, or was it months, after the day 1 problems?
ibi