You can use transactions to enforce consistency between collections that contain duplicated data. Transactions update multiple collections in a single atomic operation.
Use transactions to enforce consistency if your application must always return up-to-date data and can tolerate potential negative performance impact during periods of heavy reads.
Transactions might not be as performant as other methods of enforcing data consistency. Read performance might be negatively impacted while a transaction is open. However, transactions ensure that the data read by the client is always current.
About this Task
To use transactions, you must connect to a replica set or sharded cluster. You cannot use transactions on standalone deployments.
Before you Begin
Review the different methods to enforce data consistency to ensure that transactions are the best approach for your application. For more information, see Data Consistency.
Steps
The following example enforces data consistency in an e-commerce
application. The example schema duplicates product information in the
products and sellers collections. This schema design optimizes
queries for both products and sellers.
When a product is updated, such as when its price changes, it is
critical that the price is consistent in the products and
sellers collections. Therefore, transactions are a reasonable method
to enforce data consistency in this application.
Create the sellers collection
use test
db.sellers.insertOne(
{
id: 456,
name: "Cool Clothes Co",
location: {
address: "21643 Andreane Shores",
state: "Ohio",
country: "United States"
},
phone: "567-555-0105",
products: [
{
name: "sweater",
price: 30
},
{
name: "t-shirt",
price: 10
},
{
name: "vest",
price: 20
}
]
}
)Configure a transaction to handle updates
Note
The following example uses a transaction in mongosh. To see
transaction examples for MongoDB drivers, see Transactions.
The following example uses a transaction to update the price of
the vest in both the products and sellers collections:
// Start a session
session = db.getMongo().startSession( { readPreference: { mode: "primary" } } );
productsCollection = session.getDatabase("test").products;
sellersCollection = session.getDatabase("test").sellers;
// Start a transaction
session.startTransaction( { readConcern: { level: "local" }, writeConcern: { w: "majority" } } );
// Operations inside the transaction
try {
productsCollection.updateOne(
{ sellerId: 456, name: "vest" },
{ $set: { price: 25 } }
);
sellersCollection.updateOne(
{ },
{ $set: { "products.$[element].price": 25 } },
{ arrayFilters: [ { "element.name": "vest" } ] }
);
} catch (error) {
// Cancel transaction on error
session.abortTransaction();
throw error;
}
// Commit the transaction using write concern set at transaction start
session.commitTransaction();
session.endSession();Results
To confirm that the price was updated and that the data is consistent,
query the products and sellers collections.
Query the Products Collection
db.products.find( { sellerId: 456, name: "vest" } )Output:
[
{
_id: ObjectId("64d506c3ddebf45734d06c58"),
sellerId: 456,
name: 'vest',
price: 25,
rating: 4.7
}
]
Query the Sellers Collection
db.sellers.find( { id: 456, "products.name": "vest" } )
Output:
[
{
_id: ObjectId("64d516d9ddebf45734d06c5a"),
id: 456,
name: 'Cool Clothes Co',
location: {
address: '21643 Andreane Shores',
state: 'Ohio',
country: 'United States'
},
phone: '567-555-0105',
products: [
{ name: 'sweater', price: 30 },
{ name: 't-shirt', price: 10 },
{ name: 'vest', price: 25 }
]
}
]
Learn More
To see other ways to enforce data consistency, see: