Rediger

Bulk operations with mssql-django

This article explains how to use Django's bulk_create and bulk_update methods with SQL Server through the mssql-django backend.

bulk_create

Use bulk_create to insert multiple records in a single database operation:

from myapp.models import Product

products = [
    Product(name="Widget A", price=9.99),
    Product(name="Widget B", price=14.99),
    Product(name="Widget C", price=19.99),
]

Product.objects.bulk_create(products)

Return rows from bulk insert

By default, bulk_create doesn't return populated primary keys when using SQL Server. To enable returning primary keys, set return_rows_bulk_insert to True in your database OPTIONS:

DATABASES = {
    "default": {
        "ENGINE": "mssql",
        "NAME": "<your-database>",
        "USER": "<your-username>",
        "PASSWORD": "<your-password>",
        "HOST": "<your-server>",
        "PORT": "1433",
        "OPTIONS": {
            "driver": "ODBC Driver 18 for SQL Server",
            "return_rows_bulk_insert": True,
        },
    },
}

Important

Set return_rows_bulk_insert to False (the default) if any target table has triggers. The OUTPUT clause used to return rows isn't compatible with tables that have INSERT triggers.

bulk_update

Use bulk_update to update multiple records in a single database operation:

from decimal import Decimal

from myapp.models import Product

products = Product.objects.filter(category="widgets")
for product in products:
    product.price = product.price * Decimal("1.10")  # 10% price increase

Product.objects.bulk_update(products, ["price"])

The default parameter

The mssql-django backend still accepts a default parameter on bulk_update for backward compatibility, but it doesn't affect generated SQL in current releases.

At least one of the result expressions in a CASE specification must be an expression other than the NULL constant.

The backend now handles all-NULL update cases without requiring default.

Note

In older versions, omitting default could trigger the SQL Server error shown in the previous example. The parameter remains accepted only to avoid breaking older call sites.

Batch size

For large datasets, use the batch_size parameter to limit the number of rows per SQL statement:

from myapp.models import Product

new_products = [Product(name=f"Widget {i}") for i in range(2000)]
Product.objects.bulk_create(new_products, batch_size=500)

products = list(Product.objects.filter(name__startswith="Widget "))
Product.objects.bulk_update(products, ["price"], batch_size=500)

Related content

  • Last updated on