What is the related_name argument useful for on ManyToManyField and ForeignKey fields? For example, given the following code, what is the effect of related_name='maps'?

class Map(db.Model):
    members = models.ManyToManyField(User, related_name='maps',
                                     verbose_name=_('members'))

The related_name attribute specifies the name of the reverse relation from the User model back to your model.

If you don’t specify a related_name, Django automatically creates one using the name of your model with the suffix _set, for instance User.map_set.all().

If you do specify, e.g. related_name=maps on the User model, User.map_set will still work, but the User.maps. syntax is obviously a bit cleaner and less clunky; so for example, if you had a user object current_user, you could use current_user.maps.all() to get all instances of your Map model that have a relation to current_user.

The Django documentation has more details.

To add to existing answer – related name is a must in case there 2 FKs in the model that point to the same table. For example in case of Bill of material

@with_author 
class BOM(models.Model): 
    name = models.CharField(max_length=200,null=True, blank=True)
    description = models.TextField(null=True, blank=True)
    tomaterial =  models.ForeignKey(Material, related_name = 'tomaterial')
    frommaterial =  models.ForeignKey(Material, related_name = 'frommaterial')
    creation_time = models.DateTimeField(auto_now_add=True, blank=True)
    quantity = models.DecimalField(max_digits=19, decimal_places=10)

So when you will have to access this data you only can use related name

 bom = material.tomaterial.all().order_by('-creation_time')

It is not working otherwise (at least I was not able to skip the usage of related name in case of 2 FK’s to the same table.)

The related_name argument is also useful if you have more complex related class names. For example, if you have a foreign key relationship:

class UserMapDataFrame(models.Model):
    user = models.ForeignKey(User) 

In order to access UserMapDataFrame objects from the related User, the default call would be User.usermapdataframe_set.all(), which it is quite difficult to read.

Using the related_name allows you to specify a simpler or more legible name to get the reverse relation. In this case, if you specify user = models.ForeignKey(User, related_name='map_data'), the call would then be User.map_data.all().

The related name parameter is actually an option. If we do not set it, Django automatically creates the other side of the relation for us. In the case of the Map model, Django would have created a map_set attribute, allowing access via m.map_set in your example(m being your class instance). The formula Django uses is the name of the model followed by the string _set. The related name parameter thus simply overrides Django’s default rather than providing new behavior.

prefetch_related use for prefetch data for Many to many and many to one relationship data. select_related is to select data from a single value relationship. Both of these are used to fetch data from their relationships from a model. For example, you build a model and a model that has a relationship with other models. When a request comes you will also query for their relationship data and Django has very good mechanisms To access data from their relationship like book.author.name but when you iterate a list of models for fetching their relationship data Django create each request for every single relationship data. To overcome this we do have prefetchd_related and selected_related