The one deciding upon the acceptance or rejection of your paper is the editor, not the reviewers. In an extreme example, you do not need to fear that your paper is rejected due to a review saying nothing but:
I recommend to reject this paper because of fish.
The editor would just blacklist the reviewer and request another one. Of course, reality is more complex, but you can assume that editors put less value in reviews that show little effort from the reviewer’s side, resort to questionable arguments or seem entirely unbalanced in their assessment. In addition, if the authors convincingly address a reviewer’s critique, this will also go into the decision.
Regarding your particular case: If the negative review had fully convinced the editor that it is impossible or very unlikely that you salvage your paper, they would have rejected it, which they obviously didn’t. Moreover, if the journal accepts many publications using this database, the editor should be very worried if he is convinced by the reviewer’s criticism of the database. It may very well be that the editor just wants you to respond to possible minor criticisms from this reviewer and give some argument that your choice of data was valid.
Thus revise the paper as best as you can, addressing as much criticism as reasonably possible (i.e., stay true to your assessment of reality and do not change your claims just because the reviewer wants you to). In the response letter, address the reviewer’s or blogposts’ criticism of the database, which you usually should be able to do. There are a few exceptions from the latter, e.g., if you are just using data from another field as a benchmark for a method (in which case the validity of the data is not that problematic for you paper anyway), if the argument is based on extensive studies (in which case you can resort to arguing that these studies shall be peer-reviewed), or if there is a considerable gap between data acquisition and evaluation in your field (in which case, it’s not your responsibility to sort out these problems). However, if you cannot respond to the criticism at all, there may be indeed something wrong with the data.