CODEWITHSUNDEEP
mysqldjangooptimization
EagleOne
EagleOne

Reputation: 238

Django query fast while rendering slow

I'm running a website with Django and MySQL and I'm trying to optimize it. For that purpose I use the django debug toolbar and I pay a particular attention to the SQL part which is, I think, the reason why the load of my page is long.

When I look at the SQL part in django debug toolbar, it says that my queries take around 90ms in total. But if I put a timer around the rendering function in python, then it takes more than 3s (which is closer to what I actually see when reloading the page)...

How comes the query is that fast compared to the rendering?

EDIT: here is the code in the template simplified to its maximum: 

<div id="content_annonces" class="ui-widget ui-corner-all">
     <table>
         {% if latest_annonces_list %}
             here goes the content
         {% else %}
             nothing found
         {% endif %}
     </table>
</div>

as you can see, the only peace of code that is supposed to be costly is the SQL query requested when calling the object latest_annonces_list. But when I profile it with django debug toolbar it says this query has an elapsed of 90ms while the rendering takes around 3s ...

And here goes the content of latest_annonces_list:

result = Annonce.objects.select_related('vehicule', 'vehicule__marque')
         .filter(vehicule__marque__in=self.marques).order_by('prix')[:10]

Models here:

class Marque(models.Model):
    name = models.CharField(db_index=True, primary_key=True, max_length=100)

class Vehicule(models.Model):
    modele     = models.CharField(primary_key=True, db_index=True, max_length=100)
    gen_modele = models.CharField(db_index=True, max_length=100)
    marque     = models.ForeignKey(Marque)
    categories = models.ManyToManyField(Categorie)
    click      = models.PositiveIntegerField(default=0)

class Annonce(models.Model):
    vehicule      = models.ForeignKey(Vehicule, db_index=True)
    porte         = models.ForeignKey(Porte, db_index=True)
    carburant     = models.ForeignKey(Carburant, db_index=True)
    bv            = models.ForeignKey(Boite, db_index=True)
    prix          = models.DecimalField(db_index=True,max_digits=8, decimal_places=2)
    km            = models.PositiveIntegerField(db_index=True)
    dpt           = models.CharField(db_index=True, max_length=50)
    annonceur     = models.ForeignKey(Annonceur, db_index=True)
    img           = models.URLField()
    url           = models.URLField(max_length=2000)
    finition      = models.CharField(max_length=500)
    cylindree     = models.CharField(max_length=500)
    moteur        = models.CharField(max_length=500)
    annee         = models.DateTimeField(u'annee vehicule', db_index=True)
    pub_date      = models.DateTimeField(u'date publication')
    last_touched  = models.DateTimeField(u'derniere modification')
    last_scan     = models.DateTimeField(u'dernier scan')
    type_ann      = models.ForeignKey(Type_Annonce, db_index=True)
    type_vendeur  = models.ForeignKey(Vendeu

r, db_index=True)

Upvotes: 2

Views: 1578

Answers (1)

Peter Kilczuk
Peter Kilczuk

Reputation: 583

90ms is still pretty much for a query executed against a local database, so I guess there are many joins.

General rule in such a situation is to reduce the ORM usage by:

Probably the quickest way to have a significant gain is either to cache the QuerySets using johnny-cache http://packages.python.org/johnny-cache/ or cache the output using {% cache ... %}

Upvotes: 1

Related Questions