In my field (information theory), conference papers only got accepted or rejected, there was no such thing as "major revisions" that included another round of review. Therefore, as a senior researcher in my group once pointed out, theoretically you could change the entire paper after acceptance and still be allowed to publish (and yes, there were several conferences where a conference paper was considered a "true" publication). If your field is the same, you can ignore any suggestions from the reviewers.
However, I personally wouldn't ignore a comment just because the paper got too long. Changes I made in such a situation usually were:
- Remove DOI and weblinks from the references
- Make the figures slightly smaller
- Find paragraphs where the last line consists only of a single word and try to reformulate that paragraph to make it shorter
- If your paper contains formulas: change some smaller (unreferenced) formulas to inline
- If all that didn't help, try to shorten the introduction (the part where the background is explained)
If you are using Latex, the effects can be amazing even with the slightest change you make. Anecdotal source: I was once working on a paper that just fit the 5 pages limit. At some point, I decided to add the word "that" (the sentence was something along the lines of "the formula we obtain by" and I wanted to change it to "the formula THAT we obtain by..."). I recompiled and suddenly had a paper that was 5 pages + 1/2 column! Adding the word "that" made the last word in the paragraph move to a new line, as a consequence, another paragraph at the bottom of that page didn't fit any more and Latex decided to move two lines over to the next page. This caused a formula on the bottom of that next page to exceed limits and so on. Taking the word "that" out again made the paper exactly 5 pages again, so I decided that keeping the page limit was way more important than one sentence sounding great instead of good ;)