Installing and running ZODB

This topic discusses some boring nitty-gritty details needed to actually run ZODB.


Installation of ZODB is pretty straightforward using Python’s packaging system. For example, using pip:

pip install ZODB

You may need additional optional packages, such as ZEO or RelStorage, depending your deployment choices.


You can set up ZODB in your application using either Python, or ZODB’s configuration language. For simple database setup, and especially for exploration, the Python APIs are sufficient.

For more complex configurations, you’ll probably find ZODB’s configuration language easier to use.

To understand database setup, it’s important to understand ZODB’s architecture. ZODB separates database functionality from storage concerns. When you create a database object, you specify a storage object for it to use, as in:

import ZODB, ZODB.FileStorage

storage = ZODB.FileStorage.FileStorage('mydata.fs')
db = ZODB.DB(storage)

So when you define a database, you’ll also define a storage. In the example above, we define a file storage and then use it to define a database.

Sometimes, storages are created through composition. For example, if we want to save space, we could layer a ZlibStorage [1] over the file storage:

import ZODB, ZODB.FileStorage, zc.zlibstorage

storage = ZODB.FileStorage.FileStorage('mydata.fs')
compressed_storage = zc.zlibstorage.ZlibStorage(storage)
db = ZODB.DB(compressed_storage)

ZlibStorage compresses database records [2].

Python configuration

To set up a database with Python, you’ll construct a storage using the storage APIs, and then pass the storage to the DB class to create a database, as shown in the examples in the previous section.

The DB class also accepts a string path name as its storage argument to automatically create a file storage. You can also pass None as the storage to automatically use a MappingStorage, which is convenient when exploring ZODB:

db = ZODB.DB(None) # Create an in-memory database.

Text configuration

ZODB supports a text-based configuration language. It uses a syntax similar to Apache configuration files. The syntax was chosen to be familiar to site administrators.

ZODB’s text configuration uses ZConfig. You can use ZConfig to create your application’s configuration, but it’s more common to include ZODB configuration strings in their own files or embedded in simpler configuration files, such as configarser files.

A database configuration string has a zodb section wrapping a storage section, as in:

  cache-size-bytes 100MB

In the example above, the mappingstorage section defines the storage used by the database.

To create a database from a string, use ZODB.config.databaseFromString():

>>> import ZODB.config
>>> db = ZODB.config.databaseFromString(snippet)

To load databases from file names or URLs, use ZODB.config.databaseFromURL().

URI-based configuration

Another database configuration option is provided by the zodburi package. See: It’s less powerful than the Python or text configuration options, but allows configuration to be reduced to a single URI and handles most cases.

Using databases: connections

Once you have a database, you need to get a database connection to do much of anything. Connections take care of loading and saving objects and manage object caches. Each connection has its own cache [3].

Getting connections

Amongst [4] the common ways of getting a connection:

The database open() method opens a connection, returning a connection object:

>>> conn =

It’s up to the application to call close() when the application is done using the connection.

If changes are made, the application commits transactions to make them permanent.


The database transaction() method returns a context manager that can be used with the python with statement to execute a block of code in a transaction:

with db.transaction() as connection: = 1

In the example above, we used as connection to get the database connection used in the variable connection.


For code that’s already running in the context of an open connection, you can get the current connection as the _p_jar attribute of some persistent object that was accessed via the connection.

Getting objects

Once you have a connection, you access objects by traversing the object graph from the root object.

The database root object is a mapping object that holds the top level objects in the database. There should only be a small number of top-level objects (often only one). You can get the root object by calling a connection’s root attribute:

>>> root = conn.root()
>>> root
{'foo': 1}
>>> root['foo']

For convenience [5], you can also get top-level objects by accessing attributes of the connection root object:


Once you have a top-level object, you use its methods, attributes, or operations to access other objects and so on to get the objects you need. Often indexing data structures like BTrees are used to make it possible to search objects in large collections.