Add better indexes

shynet/analytics/migrations/0009_auto_20200414_1008.py

@@ -0,0 +1,76 @@
+# Generated by Django 3.0.5 on 2020-04-14 14:08
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('analytics', '0008_auto_20200412_0015'),
+    ]
+
+    operations = [
+        migrations.AlterModelOptions(
+            name='hit',
+            options={'ordering': ['-start_time']},
+        ),
+        migrations.AlterModelOptions(
+            name='session',
+            options={'ordering': ['-start_time']},
+        ),
+        migrations.AlterField(
+            model_name='hit',
+            name='initial',
+            field=models.BooleanField(db_index=True, default=True),
+        ),
+        migrations.AlterField(
+            model_name='hit',
+            name='location',
+            field=models.TextField(blank=True, db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='hit',
+            name='referrer',
+            field=models.TextField(blank=True, db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='hit',
+            name='start_time',
+            field=models.DateTimeField(auto_now_add=True, db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='session',
+            name='identifier',
+            field=models.TextField(blank=True, db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='session',
+            name='ip',
+            field=models.GenericIPAddressField(db_index=True),
+        ),
+        migrations.AlterField(
+            model_name='session',
+            name='start_time',
+            field=models.DateTimeField(auto_now_add=True, db_index=True),
+        ),
+        migrations.AddIndex(
+            model_name='hit',
+            index=models.Index(fields=['session', '-start_time'], name='analytics_h_session_b2667f_idx'),
+        ),
+        migrations.AddIndex(
+            model_name='hit',
+            index=models.Index(fields=['session', 'location'], name='analytics_h_session_775f5a_idx'),
+        ),
+        migrations.AddIndex(
+            model_name='hit',
+            index=models.Index(fields=['session', 'referrer'], name='analytics_h_session_98b8bf_idx'),
+        ),
+        migrations.AddIndex(
+            model_name='session',
+            index=models.Index(fields=['service', '-start_time'], name='analytics_s_service_4b1137_idx'),
+        ),
+        migrations.AddIndex(
+            model_name='session',
+            index=models.Index(fields=['service', 'identifier'], name='analytics_s_service_82ab21_idx'),
+        ),
+    ]

shynet/analytics/models.py

@@ -13,13 +13,13 @@ def _default_uuid():
 
 class Session(models.Model):
     uuid = models.UUIDField(default=_default_uuid, primary_key=True)
-    service = models.ForeignKey(Service, on_delete=models.CASCADE)
+    service = models.ForeignKey(Service, on_delete=models.CASCADE, db_index=True)
 
     # Cross-session identification; optional, and provided by the service
-    identifier = models.TextField(blank=True)
+    identifier = models.TextField(blank=True, db_index=True)
 
     # Time
-    start_time = models.DateTimeField(auto_now_add=True)
+    start_time = models.DateTimeField(auto_now_add=True, db_index=True)
     last_seen = models.DateTimeField(auto_now_add=True)
 
     # Core request information
@@ -38,7 +38,7 @@ class Session(models.Model):
         default="OTHER",
     )
     os = models.TextField()
-    ip = models.GenericIPAddressField()
+    ip = models.GenericIPAddressField(db_index=True)
 
     # GeoIP data
     asn = models.TextField(blank=True)
@@ -47,6 +47,13 @@ class Session(models.Model):
     latitude = models.FloatField(null=True)
     time_zone = models.TextField(blank=True)
 
+    class Meta:
+        ordering = ["-start_time"]
+        indexes = [
+            models.Index(fields=["service", "-start_time"]),
+            models.Index(fields=["service", "identifier"]),
+        ]
+
     @property
     def is_currently_active(self):
         return timezone.now() - self.last_seen < timezone.timedelta(seconds=10)
@@ -57,20 +64,28 @@ class Session(models.Model):
 
 
 class Hit(models.Model):
-    session = models.ForeignKey(Session, on_delete=models.CASCADE)
+    session = models.ForeignKey(Session, on_delete=models.CASCADE, db_index=True)
-    initial = models.BooleanField(default=True)
+    initial = models.BooleanField(default=True, db_index=True)
 
     # Base request information
-    start_time = models.DateTimeField(auto_now_add=True)
+    start_time = models.DateTimeField(auto_now_add=True, db_index=True)
     last_seen = models.DateTimeField(auto_now_add=True)
     heartbeats = models.IntegerField(default=0)
     tracker = models.TextField()  # Tracking pixel or JS
 
     # Advanced page information
-    location = models.TextField(blank=True)
+    location = models.TextField(blank=True, db_index=True)
-    referrer = models.TextField(blank=True)
+    referrer = models.TextField(blank=True, db_index=True)
     load_time = models.FloatField(null=True)
 
+    class Meta:
+        ordering = ["-start_time"]
+        indexes = [
+            models.Index(fields=["session", "-start_time"]),
+            models.Index(fields=["session", "location"]),
+            models.Index(fields=["session", "referrer"]),
+        ]
+
     @property
     def duration(self):
         return self.last_seen - self.start_time