It depends on the goal. If review is just review, set the topic or question, find the literature - and then write briefly.
More often we need literature survey for our article. Then you should collect those articles that are relevant for a particular question of study and which you wish to cite.
The styles of literature reviews differ a lot across the sciences. It is well accepted in mathematics to survey just 1-2 relevant article, or simply to write: "Poincare has set this problem in 18xx" - and then continue with your formulae and text. But you have to be sure that nobody has addressed this problem of Poincare before.
In economic literature this will not work. You should try to cite at least 10-20 articles, even if their link to your study is just few %. More often it is the case that you need to refer to 30-40 articles. Because literature is very dense in some fields.
The best way to learn to write is to read material of the type you wish to write, paying particular attention to the writing method and style.
As others have already mentioned, literature reviews could look different, depending on your purpose and the subject area/discipline.
My suggestion is that you look at the introductions of those papers you are interested in reviewing, and study the format of the writing from the very beginning, right up to the Method heading. If you are writing a thesis, you might also (not instead) check out a couple of theses in your field to see how literature is treated.
Also, there are many resources (books and internet materials) about how to write academically. I highly recommend these too. However, be sure that the material you rely on is reputable/reliable.
I suggest starting from the goal and work your way backward. E.g. leading the reading to the desired research gap you identified.
As the others said, articles in the 10 year range should be the main target, unless it is a truely outstanding source that is older but still a reference point.
I would disagree that only the articles of last 10 years are relevant. This is true only if you study rather new question and there was a lot written about it in the last 10 years. It is also easier to find those articles online.
But I often find an interesting area of research that has started its development about 100 years ago or even earlier, but then this path was forgotten, and there are no (or few) relevant publications in the last 10 years. For example, this is the case in economics with transport costs linear in distance. I found a lot of literature from 1930s till 1970s, but then not much.
In some cases it is almost impossible to find literature, because nobody is addressing problem the way you consider to be correct. Then I have to cite myself in half of the cases. For example, the notion of population density is practically unused in theoretical models in economics. Still my paper creates a lot of interest: https://www.researchgate.net/publication/271497718_Socio-economic_influences_of_population_density
Article Socio-economic influences of population density