This page covers various best practices that should be considered when developing applications against the Google Ad Manager API.
- Reuse service clients/stubs during the course of an execution
- Use paging when fetching objects
- Batch update requests
- Store the Ad Manager API client object when appropriate
- Use bind parameters in PQL
- Grant user privileges sparingly
Reuse service clients/stubs during the course of an execution
Creating a new service client/stub has a marginal cost associated with fetching the WSDL and allocating resources. If possible, create the service client/stub once at the beginning of an execution and make it available to classes and functions as needed.
Use paging when fetching objects
All of the services support a
get*ByStatement() method, which
allows for the filtering of results using PQL
OFFSET clauses can be used to
split large result sets into pages, preventing time outs and keeping the
response page sizes reasonable. The suggested page size is 200-500, depending
on the complexity of your objects.
Batch update requests
When changing multiple objects of the same type, you can get better
performance by sending all of the objects in the same
request. There is a marginal overhead on the client and the server for each
request, and batching can be an effective means of reducing the number of
requests. For example, use
to update a batch of Orders
rather than a single order in each call.
Use bind parameters in PQL
Bind parameters are a way to put variables into a PQL query statement. PQL
refers to a bind variable by a name without spaces starting with a colon,
:name. A code example is provided in the PQL syntax page.
We recommend using bind variables because they improve code readability by removing the need to concatenate strings and variables into a query statement. They also make it easy to reuse PQL statements, since new queries can be made by substituting the bind parameter values.
Grant user privileges sparingly
When using UserService to create/update user roles, be careful not to grant users privileges they do not need. Many features of the API can be accessed with a combination of roles rather than assigning the user an administrator role. Please refer to the permissions documentation when deciding which roles to assign a user. Also, as a third party application developer, make sure to determine what level of access your application needs when asking a network to create a user for you; a role with fewer privileges than administrator may be enough.