I give you examples from Chemistry.
-Web of Science
-Scopus/Science Direct/Springer/Taylor & Francis etc.
-CAS Sci-Finder
-Scholar.google.com
Web of science is generally known for refined, more quality papers. Surely not an exhaustive database, in that sense, but it is always a good start to manage a review, as we are humans and can't read 400 articless in a day. In short, reading review papers in web of science and looking the titles of research paper will give you quite sufficient idea about the clusters of research in that field.
-Scopus etc. databases give many papers, especially downloadable papers in case your institute has a subscription, but even not so you will still see the abstract and have an idea of what they are doing. Watch out for their bias, btw, whatever sorting option you have done they will still climb some papers above the others on the list.
-Science finder is, the real exhaustive database. You can even find many conference proceedings you will never ever have a chance to see in other places. I, for instance, once looked for the oxidative degradation of HBCD chemical, which was said to be not studied yet at that time. I have found 2-3 papers/conferences there about its advanced oxidation degradation. The best thing about this database, for chemists of course, drawing the molecule and find papers related to that molecule, instead of trying different names of the compounds.
-Scholar.google is not like a google, it is more structured, you can export citations and see many other details about papers. In many ways, you can even find it much better than other databases.
Keywords are a good start yet you can overlook many papers with that tactic. You should be more concerned about the journals in your field, much easier and systematic review, as they even do it in papers nowadays like " we have selected 51 journals and for last 10 years, reviewed such and such...". Otherwise, you will really overlook the papers did exactly what you are up to but just used other words than you might expect.
This is what it looks like, be systematic and recognize your field's journals. Things will be straightforward, then.